{"meta":{"section":"papers","title":"AI papers of the day","updated":"2026-11-14T01:32:20.996Z","stale":false,"headline":{"value":"283","label":"AI papers submitted on Nov 11, the newest full day arXiv has announced","detail":"Most common topic: Agents","count":"283","spark":[588.0,541.0,541.0,413.0,192.0,209.0,518.0,465.0,401.0,410.0,357.0,175.0,168.0,473.0,446.0,249.0,443.0,355.0,264.0,269.0,482.0,425.0,369.0,388.0,372.0,251.0,201.0,431.0,353.0,283.0],"spark_kind":"bar"},"generated":"2026-11-14T01:32:20.996Z"},"topics":["Agents","Safety and alignment","Robotics and RL","Vision and multimodal","Efficiency","Reasoning","Language models","Other"],"papers":[{"id":"2611.17334","title":"Unbundling Multimodal Vision Encoding","summary":"How multimodal language models should encode visual input remains up for debate. Most models use a pretrained vision encoder, while some recent recipes train the encoder from scratch with the language model or remove it entirely. Vision encoders bundle together three design choices: a visual prior from upstream pretraining, dedicated vision capacity, and pre-fusion placement of that capacity (i.e., before the multimodal backbone). Because published models also differ in data, architecture, and compute, comparing them cannot isolate the effect of each choice. We disentangle these choices via a controlled framework where we pretrain all models from scratch on common data and compare them on visual question answering and language benchmarks at matched lifetime compute. We find that the visual prior is not necessary; an encoder trained jointly from scratch is on par with a pretrained one, without a separate training stage. What matters instead is dedicated vision capacity: an encoder-free model trails on visual understanding even when its backbone is enlarged by the encoder's parameter count. Yet that capacity need not precede multimodal fusion, as vision-only modules added in-fusion to early backbone layers perform comparably to a pre-fusion encoder and can lower time to first token. These results also hold across backbone sizes, when encoder pretraining is treated as free, and when starting from a pretrained language model. Overall, our findings favor dedicated vision capacity trained jointly with the backbone and placed early within it.","authors":["Ellis Brown","Chris Hoang","Shengyi Qian","Shang-Wen Li","Mike Lewis","Marjan Ghazvininejad","Luke Zettlemoyer","Lucas Beyer","Mengye Ren","Jakob Verbeek","Saining Xie","Hu Xu"],"cats":["cs.CV","cs.LG"],"primary":"cs.CV","published":"2026-11-12T18:59:54Z","topic":3},{"id":"2611.17332","title":"ArchSpace: Recording, Reviewing, and Referencing LLM Architecture Research","summary":"Model architecture shapes the scaling behavior, training efficiency, and inference cost of LLMs, making it a central focus of both industry and academia. In industry, teams such as those behind DeepSeek, Qwen, and Kimi routinely introduce architectural innovations, yet public reports rarely document which ideas succeeded, which were abandoned, and why. In academia, limited compute and inconsistent baselines often restrict evaluations to small-scale, single-point comparisons, making it difficult to establish whether improvements hold at larger scales or translate into broader impact. ArchSpace addresses both gaps as a public platform for Recording, Reviewing, and Referencing Architecture Researc}: each proposal is recorded with its experiments and reasoning, reviewed by the community, and referenced by later work. To evaluate ideas before costly large-scale training, ArchSpace leverages scaling laws to predict model loss at larger scales from smaller experiments, a practice widely adopted in industry. We provide initial references that make the platform usable: Qwen3.5 and OLMo3 baselines, fitted scaling laws, and a reusable training recipe. Ideas validated through this review can in turn become new baselines, and ArchSpace thereby aims to broaden the range of ideas that shape the next generation of models.","authors":["Jie Wang","Tong Zhu","JiaHao Kuang","Zhe Xu","Yunkai Chai","Dehong Zheng","Yu Cheng","Tao Ji","ZhouHan Lin","Qi Zhang","Bowen Zhou"],"cats":["cs.CL","cs.LG"],"primary":"cs.CL","published":"2026-11-12T18:59:52Z","topic":6},{"id":"2611.17329","title":"Internalizing Investment Mandates: A Neurosymbolic RL Approach to Portfolio Allocation","summary":"Institutional portfolios operate under investment policy statements that specify asset allocation constraints, often conditioned on market regime information. Deep reinforcement learning has been shown to optimize financial performance but typically does not enforce rule compliance, making it unsuitable for real-world deployment. Existing neurosymbolic methods such as Probabilistic Logic Policy Gradient (PLPG) offer constraint-satisfying reinforcement learning (RL) via differentiable symbolic supervision but assume a finite, discrete action space. We present a PLPG-inspired continuous allocation method that estimates the probability that the policy's own executed action satisfies the investment mandate, using reparameterized Monte Carlo sampling, and adds it to the training objective. No post-hoc enforcement mechanism intercepts actions at inference time and compliance is learned by the agent. Audited directly on deterministic mean actions with no correction, the policy reduces violations of a six-rule regime-conditional mandate from 96-98% for unconstrained RL and 33.4% for a reward-penalty baseline to 4.8%, showing that a continuous-action neurosymbolic policy can substantially internalize a symbolic mandate.","authors":["Zeynep Yıldız","Satchit Chatterji","Ercan Atam","Erman Acar"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T18:59:37Z","topic":2},{"id":"2611.17323","title":"AgentSSL: Can MLE Agents Leverage Unlabeled Data?","summary":"Applying machine learning systems to new domains introduces bottlenecks in both labeled data acquisition and manual pipeline engineering. While recent MLE agents show promise in automating pipeline design, they have been studied almost exclusively in fully supervised settings, leaving them untested in realistic regimes where labeled data is scarce but unlabeled data is abundant. In this work, we introduce Agentic Semi-Supervised Learning (AgentSSL), framing SSL as a program search problem and studying how agents navigate this space under limited supervision. We find that even under weak feedback and matched compute, agents can synthesize code that combines and adapts established SSL techniques, exceeding the strongest human-engineered baselines in 8 of 12 settings by up to +21.8%. However, our results identify program selection as a critical bottleneck in AgentSSL, where performance drops by as much as 47% when selecting based on the agent's feedback metric. To systematically study this bottleneck, we introduce ProgramSelectBench, a large-scale benchmark with a combined total of 16,977 agent-generated and human-engineered SSL programs, and evaluate both unsupervised and active program selection methods. We further evaluate AgentSSL on real-world ecology detection tasks, highlighting the opportunities and challenges of transferring agent-discovered strategies beyond controlled benchmarks.","authors":["Ethan Lin","Akanksha Sarkar","Ziang Liu","Kristin Branson","Jennifer J. Sun"],"cats":["cs.AI","cs.LG"],"primary":"cs.AI","published":"2026-11-12T18:59:06Z","topic":0},{"id":"2611.17320","title":"Learning When Diffusion Language Models Should Regenerate","summary":"Remasking allows diffusion language models to revisit generated text, but effective correction requires knowing when and what to revise. Our diagnosis on Sudoku reveals two gaps. Hidden states contain error information that token confidence does not reliably expose, and remasking an isolated error usually reproduces it when the surrounding context remains fixed. These findings motivate DiVR , a verifier that predicts whether a partially generated sequence will reach a correct answer. Trained on the final correctness of the model's own trajectories, DiVR decides when to regenerate a block and which alternative to retain, while the generator itself stays unchanged. On five reasoning benchmarks with LLaDA-8B-Base, DiVR raises the average accuracy of the fine-tuned generator from 35.3\\% to 43.4\\% at $1.9$-$3.5\\times$ the cost of greedy decoding, above search baselines that use $3.8$-$7.4\\times$. It also complements post-training such as supervised fine-tuning and reinforcement learning, and generalizes to other generators, including Dream.","authors":["Minhae Oh","Jeonghye Kim","Jinwoo Koo","Yeongoon Kim","Jimin Seo","Minsik Seol","Jungwoo Lee"],"cats":["cs.CL"],"primary":"cs.CL","published":"2026-11-12T18:58:50Z","topic":3},{"id":"2611.17316","title":"CoReMem: Cost-Internalizing Memory for Successive Coding Tasks with Prospective Valuation","summary":"Coding agents tackle successive tasks on an evolving repository, carrying forward edits and writing memory before the next task is known. Selecting memory for unknown next tasks requires balancing future usefulness against the costs of writing and reading it. The actor must also decide when acting on available memory saves effort as the task unfolds. We introduce Coding-Repository Memory (CoReMem), a cost-internalizing framework that guides memory design and actor adaptation through measured memory value on the next task. Prospective segment valuation compares runs of the next task with and without individual trajectory segments from the same inherited repository state, guiding a compact memory built around anchors: repository locations and executable checks. Actor adaptation estimates on-policy memory value within rollout groups with and without memory, using a reward that prioritizes resolution and penalizes step count in successful runs. Provenance-gated value credit assigns the additional value term to steps identified as using memory-supplied anchors, while every step retains outcome credit. On ChainSWE, CoReMem raises the 27B actor's resolve rate from 21.0% to 25.7%, with 17% fewer steps and 11% fewer total cache-aware tokens, including memory-module calls, relative to the untrained actor model without memory.","authors":["Peiwen Li","Harshit Verma","Sophia Zhou","Haiwen Wang","Ngoc Bui","Heng Chang","Fei Fei","David van Dijk","Rex Ying"],"cats":["cs.AI","cs.CL","cs.SE"],"primary":"cs.AI","published":"2026-11-12T18:58:22Z","topic":0},{"id":"2611.17314","title":"Linking Rapid Learning to Compositional Structure in Neural Networks: Evidence from Additive Analogies in Meta-Learned Parameter Spaces","summary":"Humans can learn rapidly from limited experience, while neural networks typically require extensive training. In this work, we investigate what properties of a neural network can enable it to learn rapidly. We study neural networks that have been optimized via meta-learning, a technique that exposes a network to many related functions to facilitate its subsequent learning of similar functions. Prior work has shown that these meta-learned networks can learn new functions rapidly, but how they do so remains unclear. To investigate their internal structure, we take inspiration from classic work demonstrating additive relationships among the learned vectors that neural networks use to represent words (e.g., \"king\" minus \"man\" plus \"woman\" approximates \"queen\"). Analogously, we look for additive relationships among a meta-learned network's \"representations\" of different functions; when a meta-learned network rapidly learns to implement a specific function, all of its adapted parameter values can be concatenated into a single vector that we treat as representing the function. We find that such parameter vectors can be added and subtracted according to symbolic relationships among the functions they represent, yielding new parameter vectors that successfully represent the functions predicted by the corresponding symbolic manipulations: the networks specified by these artificially recombined vectors actually implement the intended functions without any direct training on them. Our results suggest that meta-learning can tune a network's starting point to organize subsequent learning along systematic, functionally meaningful trajectories through the space of possible parameter vectors, providing a geometric basis for realizing inductive biases in neural networks.","authors":["Abigail L. Tenenbaum","R. Thomas McCoy"],"cats":["cs.LG","cs.CL"],"primary":"cs.LG","published":"2026-11-12T18:57:49Z","topic":2},{"id":"2611.17313","title":"Sketching Language Models for Context Attribution","summary":"Context attribution asks how removing parts of a prompt changes a language model's output. We study leave-one-out (LOO) attribution for a prompt of $n$ tokens and a fixed measurement function. Computing all LOO values directly requires $n$ model evaluations. Under a stability assumption, our algorithm estimates all LOO values to error proportional to $\\varepsilon$ with constant probability using $O(\\sqrt{\\log(1/\\varepsilon)}/\\varepsilon)$ evaluation calls. To complement our upper bound, we prove a matching $Ω(\\sqrt{\\log(1/\\varepsilon)}/\\varepsilon)$ lower bound on the number of evaluation calls in an oracle model. Empirically, we verify coefficient decay consistent with stability and evaluate the performance of our sketching algorithm on a small language-model retrieval task. Technically, our result is based on a sketching algorithm for arithmetic circuits whose Taylor coefficients satisfy the stability condition recently introduced by Gunn [Gun26]. A direct adaptation of their forward differentiation sketches would require $\\widetilde{O}(\\log(n)/\\varepsilon^2)$ evaluation calls. Our key observation is that we can apply forward differentiation to the backpropagation circuit, with only constant overhead over a forward sketch of the same Taylor degree. This reduces the variance of the sketches and is key to our improved evaluation bound. Our lower bound builds on the techniques of Amsel, Chen, Duman Keles, Halikias, Musco, and Musco [ACDH+26, SIAM Journal on Matrix Analysis and Applications] for lower bounding the number of matrix-vector queries to estimate the diagonal elements of random matrices.","authors":["Christopher Liaw","Aranyak Mehta","Grigoris Velegkas","Felix Zhou"],"cats":["cs.DS","cs.AI","cs.LG","stat.ML"],"primary":"cs.DS","published":"2026-11-12T18:57:48Z","topic":6},{"id":"2611.17309","title":"Classical Energies Learn to Reason 1,000x Faster than Looped Models","summary":"Looped models have emerged as a prominent paradigm for reasoning, with successes on simple reasoning puzzles such as Sudoku-Extreme and Maze-Unique. However, these models can be unstable, slow to train, and use many parameters and FLOPs to solve simple puzzles. Descent on an energy not only mitigates these issues but also offers advantages: built-in early stopping, compute that adapts to problem difficulty, fast training without differentiating through a solver, and interpretable weights that learn the constraints of the puzzle. Here we demonstrate this with a learned version of Hopfield & Tank's energy from the 80s, which we term CRAM, solves 99.996% of Sudoku-Extreme and 100% of Maze-Unique test puzzles after 42 ms and 132 ms of training on one GPU, at least 1,000x faster than looped models on these tasks. CRAM also matches baseline looped models at their highest accuracy on Sudoku-Extreme using at least 282x fewer inference FLOPs. Our work suggests two things. First, that energies are an extremely promising direction for training reasoning from data. Second, a caution against using trivially solvable benchmarks like Sudoku and Maze to test the scalability of looped reasoning techniques.","authors":["Benjamin Hoover","Hugh Blayney","James C. R. Whittington"],"cats":["cs.AI","cs.LG"],"primary":"cs.AI","published":"2026-11-12T18:56:56Z","topic":1},{"id":"2611.17308","title":"Quantifying structural stability of transparent models for general insurance pricing","summary":"The short duration of general insurance contracts leads to rapid distributional shifts in the data, warranting regular model updates. This process may yield a post-update model that differs considerably from the pre-update model in both predictions and structure. Structural stability is upstream of predictive stability, yet existing structural stability metrics are mostly restricted to simple decision trees. Thus, we first introduce the notion of functional transparency, characterising models with a prediction function that can be expressed as a sum of univariate functions for main effects and bivariate functions for pairwise interactions. Then, we propose the Mean Squared Shape function distance, a metric to quantify the structural stability of functionally transparent models, which are particularly relevant in actuarial science. Using three public auto insurance datasets, we present a benchmark study to demonstrate the usefulness of our metric in identifying a stable and accurate transparent model. Generalised linear models with centred numerical covariates are stable, and explainable boosting machines (EBMs) offer a compromise between stability and performance. We also find that stabilisation strategies, such as sequential modelling and fixed hyperparameters, can substantially improve stability for both EBM and XGBoost models, with negligible impact on performance.","authors":["Félix Côté","Marie-Pier Côté"],"cats":["stat.ML","cs.LG"],"primary":"stat.ML","published":"2026-11-12T18:56:54Z","topic":7},{"id":"2611.17303","title":"Towards a Theory of Learnability via Debate","summary":"Debate has been recently proposed as a training protocol that is hoped to produce more aligned models even as model capabilities scale. The idea is that the human, instead of providing supervision in the form of labels, judges a debate between two models that compete to prove the correctness of opposite labels. It can be shown that functions of a much higher computational complexity can be correctly judged in this way, compared to the complexity of functions that can be directly evaluated using a computationally weak evaluator. This suggests that perhaps debate can be used to teach very complicated functions to models. However, previous work has largely ignored the fact that this is a training protocol, and simply analyzes the game being played between the two debaters for a known judge. We take the first steps towards understanding the learning problem implicit within the protocol, by formulating it as an online learning problem. We show that once we formulate the learning problem precisely, there are cases where the best way to learn given access to a judge is different from running an adversarial debate. This motivates us to define a cooperative version of the protocol, whose learnability we characterize using a combinatorial dimension. We also consider the possibility of the human judge being fallible and occasionally judging the debate incorrectly. We show that a small number of errors in judgement can lead to a large error in learning.","authors":["Tosca Lechner","Vinayak Pathak","Sivan Sabato","Ruth Urner"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T18:55:58Z","topic":1},{"id":"2611.17302","title":"FLEX-LBP: Learning Local Binary Patterns with Exact Binary Inference for Texture Classification","summary":"Local Binary Patterns (LBP) remain among the most widely used texture descriptors, from industrial inspection to medical imaging. Yet their radius, sampling angles, color transform and thresholds are fixed by hand for each dataset. We present FLEX-LBP, which learns these parameters by backpropagation while every forward pass computes exact binary codes and hard histogram counts. A straight-through estimator acts only in the backward pass, so the deployed descriptor is an interpretable hard-binary LBP with 90 learnable parameters and unchanged inference cost. In a controlled ablation on six color texture benchmarks, the component worth learning is dataset-dependent. One configuration, applied unchanged everywhere, learns the geometry and thresholds with the color transform frozen. It improves the nearest-neighbor accuracy of the identical fixed descriptor by 3.55 points on average over five seeds, significantly on five of six benchmarks and never significantly worse. With a linear discriminant readout, it gains 0.9 points; the fully learned descriptor gains 2.7 and has the highest mean. FLEX-LBP is more accurate than five classical LBP-family descriptors, PCANet and a neural LBP on five of six benchmarks. Its mean nearest-neighbor accuracy exceeds that of frozen ImageNet ResNet-50 features (23.5 million parameters). On the Kather colorectal cancer histology benchmark, the fully learned descriptor reaches 93.8%, significantly above the best classical descriptor (Completed LBP, 92.8%). The learned radii adapt in directions that recur across seeds. Noise drowns single-pixel comparisons and illuminant change flips inter-channel ones; median sampling or gray-world normalization restores 96-97%. The code is available as a drop-in PyTorch layer at https://github.com/medalimoussa/FLEX-LBP.","authors":["Mohamed Alimoussa","Amal El Kaid"],"cats":["cs.CV","cs.LG"],"primary":"cs.CV","published":"2026-11-12T18:55:56Z","topic":1},{"id":"2611.17298","title":"Consistent Video-to-Video Translation via Explicit Correspondences","summary":"Interactive video-to-video applications require real-time generation while maintaining long-range temporal consistency. However, recent methods achieve speed by restricting attention to a fixed-length causal window. This limits long-range consistency whenever relevant history falls outside the context window. To address this limitation, we introduce vid2vid-long, a correspondence-based approach grounded in a simple observation. Video-to-video translation provides an explicit cross-frame reference: the regions that match in the input video should match in the output video too. Concretely, we compute patch-level correspondences in the input, guiding retrieval of previously generated outputs in the corresponding regions, and condition the model on these patches through correspondence cross-attention. Our experiments demonstrate that vid2vid-long consistently improves long-range temporal consistency across recent video models while preserving generation quality and real-time performance.","authors":["Gaurav Parmar","Zhengqi Li","Richard Zhang","Jun-Yan Zhu","Srinivasa Narasimhan","Eli Shechtman","Yotam Nitzan"],"cats":["cs.CV","cs.AI","cs.GR","cs.LG"],"primary":"cs.CV","published":"2026-11-12T18:55:09Z","topic":3},{"id":"2611.17297","title":"Goal Encoding in LLMs","summary":"We investigate whether and how LLMs encode goals: internal neural representations that precede and guide action, and that are flexible enough to be realized through many different execution paths. We construct a coding setting with multiple sub-tasks that allows us to evaluate goal-seeking behavior and the selection of sub-goals. In this setting, we localize an encoding of the next sub-goal in the outputs of a sparse set of attention heads. The same heads can be used to decode the LM's future actions in natural language (e.g., \"I'm checking if a word is a palindrome\"). We verify that this encoding is causal for next action and is plastic: transplanting it into a different environment, e.g., another programming language, triggers the same operation realized with a different sequence of tokens. The LM also sustains the encoding throughout the action's execution, which is puzzling because a transformer LM without recurrence cannot carry a state forward for more tokens than it has layers left. We propose that it does not need to. Under a non-recurrent goal-seeking (NRGS) account, the LM frequently re-reads the goal from the task specification, compares it with the progress visible in the growing context, and pursues what remains. We test the predictions of this account with targeted causal interventions.","authors":["Arnab Sen Sharma","Eric Todd","Sheridan Feucht","Andy Arditi","Rohit Gandikota","Natalie Shapira","Gabriele Sarti","Constanza Fierro","Lisa Schut","David Bau"],"cats":["cs.CL","cs.AI","cs.LG"],"primary":"cs.CL","published":"2026-11-12T18:55:06Z","topic":6},{"id":"2611.17294","title":"Metal Sandbox: Isolated Execution and Command Scheduling for GPU Coding Agents on Apple Silicon","summary":"On-device inference requires kernels that make effective use of local accelerators. Efficient training and evaluation of models that generate Apple Metal kernels benefit from native macOS tools and concurrent agent sessions. These sessions alternate between model requests, development commands, and GPU execution, leaving GPU capacity idle for much of a rollout. We present Metal Sandbox, infrastructure that combines restricted per-instance file access, native command execution, and command-level scheduling across interconnected Macs. Each agent retains its workspace and private development environment. Parallel commands can run concurrently, while serial commands share one execution slot per node; eligible waiting commands can migrate to available Macs and return results to the original agent. Integration with Harbor supports task execution, automatic verification, and rollout collection through a common workflow. Controlled experiments on four Mac mini M4 nodes achieve $1.68\\times$--$2.05\\times$ gains in command throughput under concentrated demand. Sixteen static agent traces give an estimated $2.14\\times$ sustained-throughput gain for four sessions per node versus one, assuming continuously replenished sessions and unchanged mean durations. Replaying these measured traces under dynamic scheduling reduces mean tool waiting by 18.8% and batch elapsed time by 22.3%, assuming unchanged phase durations. Metal Sandbox and its Harbor integration are open-source on GitHub.","authors":["Haozhe Wang","Lina Chen","Fanqi Wan","Qi Han","Yinmin Zhang","Zheng Ge","Daxin Jiang","Junchi Yan"],"cats":["cs.LG","cs.AI","cs.CL","cs.DC"],"primary":"cs.LG","published":"2026-11-12T18:54:42Z","topic":0},{"id":"2611.17287","title":"Negative-Only Policy Optimization for One-Sided Verifiable Rewards","summary":"Reinforcement learning with verifiable rewards (RLVR) typically relies on a verifier that can reliably judge whether sampled solutions are correct. However, in many realistic settings, complete verification is generally unavailable, while cheap checks can only falsify failures. We formalize this regime as one-sided verifiability (OSV): verified negatives are reliable counterexamples, but verified positives are merely non-falsified candidates and may contain many false positives. This asymmetry makes standard RLVR unreliable, because reinforcing verified positives can directly reward shortcuts that pass partial checks. To address this issue, we propose Negative-Only Policy Optimization (NOPO), which treats verified positives as unlabeled and updates the policy only by suppressing trusted verified negatives. Specifically, NOPO combines a reference-relative negative suppression loss with a group-level adapter that scales each update by the proportion of non-falsified responses. Across unit-test, constraint-check, and LLM self-verifiers, NOPO improves pass@k over baselines on code, logic, and math benchmarks when oracle rewards are unavailable. These gains persist under inference-time scaling and cross-benchmark transfer.","authors":["Jiacheng Xu","Shuo He","Fuxiang Zhang","Chaojie Wang","Bo An"],"cats":["cs.LG","cs.CL"],"primary":"cs.LG","published":"2026-11-12T18:52:33Z","topic":2},{"id":"2611.17280","title":"Synthesis Smuggling: Maliciously Hiding Inferred Personal Attributes in LLM Outputs","summary":"We propose Synthesis Smuggling, a novel threat model in which an adversarial LLM infers sensitive information about a user from routine assistance-seeking interactions and steganographically encodes that into innocuous-looking model-generated content. The user may then unknowingly transmit that information by sharing the generated artifact, say as an email, social-media post, or message, through ordinary communication channels. An adversary monitoring these channels can discover this information. We demonstrate this threat by training two open-weight models to infer gender and age-group labels from conversational cues and encode their predictions using invisible characters or word choices. These results show that user-shared model outputs can become a covert channel for personal inferences, even when the model itself has no direct means of contacting an adversary.","authors":["Gautam Galada","Supreeth Shastri","Sagnik Ray Choudhury"],"cats":["cs.CL","cs.AI"],"primary":"cs.CL","published":"2026-11-12T18:51:05Z","topic":6},{"id":"2611.17267","title":"LiveGeo: Benchmarking Geographic Agents in Continuously Evolving Worlds","summary":"We introduce GeoWorld, a controllable and continuously evolving geographic environment platform for LLM agents, and LiveGeo, a benchmark for evaluating reliable decision-making in real-world geographic environments using GeoWorld. Unlike existing evaluations that largely reason and plan over fixed geographic states, GeoWorld is grounded in real-world geographic structure and allows external world states to evolve along their own temporal processes while preserving verifiable state histories. LiveGeo places geographic tasks within this evolving environment, requiring agents to actively acquire task-relevant information, verify whether the evidence supporting prior decisions remains valid, and revise their plans when necessary. Our results reveal that even frontier LLMs struggle with static tasks, with no model consistently excelling across tasks, while strong static performance does not ensure dynamic reliability. Further analysis identifies a key bottleneck: agents struggle to acquire and translate relevant evidence into valid decisions and maintain their correctness as the world evolves. By releasing LiveGeo together with its GeoWorld platform, we aim to provide the community with a flexible and extensible infrastructure for studying the next generation of agents capable of reliable action in evolving geographic worlds.","authors":["Pengcheng Zhou","Xuanyu Liu","Yanchen Yin","Tianjie Ju","Bobo Li","Shengqiong Wu","Hao Fei","Mong-Li Lee","Wynne Hsu"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T18:45:58Z","topic":0},{"id":"2611.17246","title":"Programming Quantum Circuit Parameters with Large Language Models","summary":"Large language models (LLMs) have recently demonstrated the ability to generate parameters for variational quantum circuits without conventional optimization. However, the conditions under which this approach is effective remain unclear. Here, we investigate a frozen, general-purpose LLM that generates complete circuit parameter vectors from symbolic circuit and Hamiltonian descriptions, with optional iterative energy feedback. We evaluate the approach on general random Hamiltonians, structured parent Hamiltonians, and molecular systems. For general Hamiltonians, LLM-generated parameters exhibit larger energy errors than conventional Adam optimization, while additional feedback generally provides limited improvement. In contrast, for structured Hamiltonians, the LLM can generate promising low-energy parameter configurations, suggesting that identifiable physical and mathematical structure facilitates parameter inference. We also show that direct LLM predictions of minimum energies can violate the Hamiltonian spectral lower bound, whereas generated circuit parameters allow independent evaluation of physically realizable energy expectation values. Finally, we investigate blind circuit interpretation and find that the LLM can predict measurement outcomes even when it fails to identify the underlying quantum algorithm. These findings highlight the role of problem structure, the limitations of iterative LLM optimization, and the importance of independently verifiable predictions in quantum parameter programming.","authors":["Tiago de Souza Farias","Caio Moraes Porto","Matheus da Silva Fonseca","Nicolas Armando Cabrera Carpi","Anderson de Araujo Oliveira","Celso Jorge Villas-Boas"],"cats":["quant-ph","cs.LG"],"primary":"quant-ph","published":"2026-11-12T18:34:23Z","topic":6},{"id":"2611.17241","title":"Ballad: Bandit-Based LLM Routing for Automated Heuristic Discovery","summary":"Modern automated heuristic discovery (AHD) systems increasingly rival hand-crafted heuristics by leveraging Large Language Models (LLMs) to iteratively mutate and refine solutions. A natural assumption is that more powerful and more expensive models consistently yield better heuristics. We challenge this empirically: across multiple NP-hard combinatorial optimization problems, we find no consistent advantage for larger, costlier models. This raises a deeper question: given a pool of LLMs spanning frontier and freely hosted or open-weight models, how should one adaptively select which model to invoke at each step of the search? The challenge is non-trivial. The optimal LLM may shift over the course of a run, and because one LLM's output seeds the context for subsequent calls, each model's reward signal is non-stationary and entangled across the heuristic population, thus violating the independence assumptions of classical online decision-making settings. We address this through Ballad (Bandit-based LLM Routing for Automated Heuristic Discovery), an online bandit algorithm that learns to orchestrate a pool of LLMs into an ensemble that matches or exceeds what any soloist LLM achieves alone. This orchestration is powered by an LLM selection policy that favors rare, high-scoring heuristics over stable averages, a DAG-based credit propagation mechanism through the genealogy of the heuristic population, and discounting of stale observations to remain attuned to each model's current utility. Across five NP-hard combinatorial optimization problems and a pool of four models of varying capability and cost, Ballad not only matches the best individual LLM in hindsight but, in several cases, surpasses every individual model while reducing API cost by up to 65.6% relative to always using the strongest model.","authors":["Samidha Verma","Ankit Anand","Sayan Ranu"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T18:28:08Z","topic":2},{"id":"2611.17239","title":"When Both Sides Adapt: The Challenges of Evaluating Interactive LLM Systems","summary":"In interactive systems where users engage with LLMs, users are adaptive and, after observing LLM responses, can change their prompting strategy, as well as the type and difficulty of subsequent tasks. Thus model responses can influence the set of prompts on which the model is subsequently evaluated. We study the implications of this user adaptivity and prompt endogeneity for the evaluation of LLMs. We define two classes of estimands for comparing models: the common-distribution effect, which compares model performance on the same distribution of prompts, and the induced-distribution effect, which compares each model on the prompt distribution generated through user interaction. We show that offline benchmark evaluation and live A/B testing target different quantities, with the standard difference-in-means estimators for the two methods converging to the common-distribution effect and the induced-distribution effect, respectively. This distinction can affect decision-making: the two comparisons can rank models differently and lead to discrepancies in deployment recommendations. We characterize when such ranking reversals occur and provide a sufficient condition under which the rankings are guaranteed to agree. This result provides a practical heuristic for assessing whether offline evaluation is sufficient to reproduce the decision suggested by a live experiment. Third, we show, in a stylized linear model of interactions, that user adaptivity can increase variance in estimation. Finally, we discuss implications of user adaptivity for the interpretation of experiments and show that adaptivity creates a common mechanism leading to common evaluation challenges, such as reward hacking and novelty effects. Together, this work highlights the necessity of measuring both model performance as well as changes in the user behavior that the model induces.","authors":["Hannah Li"],"cats":["stat.ME","cs.AI"],"primary":"stat.ME","published":"2026-11-12T18:26:51Z","topic":6},{"id":"2611.17235","title":"Al-Ghasali: A Computational Framework for Simulating Islamic Jurisprudential Rulings via Finite State Automata and Large Language Models","summary":"This paper presents Al Ghasali, a computational framework designed to formalize and simulate Islamic jurisprudential (Fiqh) rulings. Building upon recent set theoretic formalizations of al sabr wa al Taqsim that extract minimal operational rules as simple Horn clause logical programs, this work proves that the semantics of closed Fiqh chapters collapse completely to deterministic Finite State Automata (FSA). We establish a formal computational hierarchy, demonstrating that restricting formalization to finite propositional Horn theories guarantees decidability and constant time (O(1)) verdict extraction after offline compilation. The framework formally proves that hierarchical compositions of sub-chapter FSAs preserve this O(1) complexity, and that temporal juristic requirements, such as Tartib (order), are deterministically flattened into static ontological states, preserving the memoryless property of the automaton. To generate human-readable juristic summaries, Al Ghasali incorporates Large Language Models (LLMs) via an architecture where the deterministic FSA strictly controls the probabilistic LLM. Crucially, the paper formulates sufficient design safety conditions for this FSA LLM integration, ensuring strict adherence to the epistemological boundaries of Usul al Fiqh. To neutralize the risk of LLM hallucination, the system employs a Grammar Guided Constrained Decoding mechanism. We formally define the mapping function that dynamically constructs unambiguous right linear grammars from the FSA execution history. The paper mathematically proves that this architecture, augmented by a Bounded Neutral Lexicon and Valence Locking, guarantees the absolute absence of Decisional Hallucination, Juristic Factual Hallucination, and Semantic Inversion reducing the LLM to a pure linguistic transformer of deterministic juristic facts","authors":["Elnaserledinellah Mahmoud Abdelwahab"],"cats":["cs.AI","cs.FL"],"primary":"cs.AI","published":"2026-11-12T18:23:56Z","topic":6},{"id":"2611.17230","title":"Two mathematical models of knowledge distillation","summary":"Many hypotheses compete to explain the successes of knowledge distillation. To help address this, we propose and analyze a mathematical model of distillation, which suggests that distillation's performance comes not from obtaining better models but from easier to optimize landscapes. For generalized linear models trained with stochastic gradient descent, we prove that distillation fits performant student models asymptotically more quickly than non-distilled models. In rank-1 matrix approximation, we characterize conditions on the target matrix under which gradient descent with distillation converges strictly faster than training on the supervised objective. The theory helps delineate the ways distillation provides benefits (i.e., in optimization speed, not in generalization), and experiments on real datasets corroborate the theoretical predictions.","authors":["Audrey Xie","Ludwig Schmidt","John Duchi"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T18:21:37Z","topic":4},{"id":"2611.17228","title":"Linear Regression with Self-Selection Bias in Polynomial Time","summary":"We revisit the problem of estimating $k$ linear regressors with self-selection bias in $n$ dimensions up to error $\\varepsilon$ with the maximum self-selection criterion, as introduced by Cherapanamjeri, Daskalakis, Ilyas, and Zampetakis [CDIZ23, STOC]. Our main result is a $\\mathrm{poly}(n,k,1/\\varepsilon)$ time algorithm for the unknown-index variant of this problem, thereby improving upon the running times of the best known algorithms due to Gaitonde and Mossel [GM24] and Kalavasis, Mehrotra, and Zhou [KMZ26, COLT] with exponential dependence on $k$. Our algorithm succeeds in the presence of correlated noise between regressors, which captures the related setting of max-affine regression. Consequently, we also give the first polynomial-time algorithm for this problem, improving on the prior works of Ghosh, Pananjady, Guntuboyina, and Ramchandran [GPGR22, IEEE Transactions on Information Theory] and Gaitonde and Mossel [GM24] with exponential dependence on $k$. Prior works exhaustively search over a $k$-dimensional subspace spanned by the regressors, using conditional moments of the response to test candidates. In contrast, our algorithm is driven by an improved understanding of the moments of the covariates. We show that, restricting to observations with large response values, the degree-$(\\log n)$ Hermite moment of the covariates approximately decomposes as a sum of tensor powers of the regressors. The first step of our algorithm can be viewed as a tensor decomposition algorithm for these implicitly represented high-order tensors, which outputs unit vectors parallel to the regressors. A secondary step based on a simple variant of ordinary least squares extracts estimates for the regressors given these unit vectors. We believe our analytic insight and algorithmic components may have applications to designing efficient algorithms for other latent-variable problems.","authors":["Felix Zhou"],"cats":["math.ST","cs.DS","cs.LG","stat.ML"],"primary":"math.ST","published":"2026-11-12T18:21:11Z","topic":1},{"id":"2611.17226","title":"Sheaf-Inspired Structured Scoring Extensions of LightGCN for Collaborative Filtering","summary":"Graph-based collaborative filtering learns user and item representations by propagating embeddings over a user--item interaction graph. We investigate a modular, sheaf-inspired scoring extension placed on top of a three-layer LightGCN backbone. The motivation comes from cellular sheaves, where local vector spaces are related by linear restriction maps; here, this idea motivates structured, pair-dependent transformations of user and item embeddings rather than a change to graph propagation. We compare the identity inner-product scorer, diagonal, rank-one low-rank, and full matrix residual scorers, and a parameter-matched residual MLP baseline (\\texttt{mlp\\_full}) on Amazon-Book, Gowalla, MovieLens-1M, and Yelp2018. We report ranking quality and catalog coverage at cutoffs 10 and 20. The selected structured variants show modest, dataset-dependent changes in ranking quality and sometimes substantial changes in coverage. Our method is inspired by sheaf geometry but does not construct a cellular sheaf or implement sheaf-Laplacian diffusion; the experiments therefore do not establish a causal benefit from sheaf geometry itself.","authors":["Ksenia Belkovskaya","Timur Lidzhiev","Maria Stebliy","Alexander Tarakanov"],"cats":["cs.IR","cs.CE","cs.CG","cs.LG"],"primary":"cs.IR","published":"2026-11-12T18:20:22Z","topic":3},{"id":"2611.17225","title":"Data Archive for 2D and 3D Electrical Impedance Tomography with moderate conductivity contrast phantoms","summary":"This document describes open Electrical Impedance Tomography (EIT) measurement data that were collected from biomedical imaging inspired phantoms with moderate conductivity contrasts. The data were collected with the KIT4 and KIT5 EIT measurement devices at the University of Eastern Finland, and are intended for EIT image reconstruction algorithm development and testing. The data from thoracic imaging inspired cylindrical and chest-shaped phantoms are suitable for 2D algorithms, and the data from an intracerebral hemorrhage imaging inspired head-shaped phantom with resistive skull are suitable for 3D algorithms. In addition to the measurement data, computational meshes and example reconstruction codes are provided and sample reconstructions included.","authors":["S. J. Hamilton","A. Hauptmann","A. Hänninen","V. Kolehmainen","J. Niskanen","T. Savolainen","J. Toivanen"],"cats":["eess.IV","cs.LG"],"primary":"eess.IV","published":"2026-11-12T18:19:41Z","topic":3},{"id":"2611.17223","title":"Combinatorial Bernoulli Factories with Polynomially Many Coins","summary":"We provide an efficient implementation of combinatorial Bernoulli factories. Such a factory turns coins of unknown biases $x$ into a random vertex of a polytope $P$ whose mean is exactly $x$. For every affine slice of the cube (the class of polytopes that admit a factory), we give a single algorithm whose expected number of coin flips and running time at every interior $x$ are polynomial in the dimension and in $1/{δ_*}$, where ${δ_*}=\\min_i\\{x_i,1-x_i\\}$ is never supplied to the algorithm. For perfect matchings this gives $O(n^{8}{δ_*}^{-2}\\log(n/{δ_*}))$ flips, while the factory of Niazadeh, Paes Leme, and Schneider [STOC 2021] spends at least $n^{2n-1}$ flips in expectation at the uniform doubly stochastic matrix. For other polytopes, our method rounds exactly given a point deep inside $P$ and the promise that $x$ is far from the boundary of $P$.","authors":["Renato Paes Leme","Rad Niazadeh"],"cats":["cs.DS","cs.GT","cs.LG","math.PR","stat.CO"],"primary":"cs.DS","published":"2026-11-12T18:19:37Z","topic":4},{"id":"2611.17220","title":"Scalable Conversational Assessment: Chat-Based Exams with a Two-Agent LLM Pipeline","summary":"Oral exams are compelling because (1) students can be asked to provide their reasoning for an answer and (2) their interactive nature permits follow-up questions to resolve ambiguous answers. But oral exams scale poorly. In this paper, we present scalable conversational assessment, an LLM-based system that conducts interactive text-based conversations with students to assess their ability to make and support an argument. We use a two-agent architecture to separate concerns and reduce the likelihood that answers are leaked to the student: the interviewer interacts directly with the student, but only receives guidance about the next question to ask the student; the hidden evaluator knows the goals of the whole activity, tracks the student's progress using a rubric, and guides the interviewer on when to follow-up on the current question or proceed to the next question. The system was first tested using a collection of simulated students with LLM personas of a particular ability level or adversarial tactic (e.g., answer-extractors, prompt-injectors, confident fabricators). A novel element of this evaluation was that it required the use of a fictional assessment topic to prevent personas that shouldn't have information from using knowledge present in the LLM's pre-training. The system was also deployed in a small, elective AI literacy course in both formative and summative contexts. Through manual and post-facto AI review of the transcripts we identified opportunities to improve the follow-up question generation and ensure that grading only credited ideas that the student (and not the interviewer) contributed.","authors":["Geoffrey Challen","Craig Zilles"],"cats":["cs.CY","cs.AI"],"primary":"cs.CY","published":"2026-11-12T18:16:37Z","topic":0},{"id":"2611.17214","title":"Introducing Feedback Delays to Non-Delayed Adversarial MDPs: Optimal $\\sqrt S$ and $\\sqrt K$ Dependencies in Tabular and Linear AMDPs","summary":"We study regret minimization in episodic Adversarial Markov Decision Processes (AMDPs) with unknown transitions and bandit feedback. In tabular AMDPs with $S$ states, $A$ actions, and horizon $H$, the best known $K$-episode regret is $\\tilde O(HS\\sqrt{AK})$, whose $\\sqrt S$ gap to the $Ω(H\\sqrt{SAK})$ lower bound has long remained open. In linear AMDPs with $d$-dimensional features, after a five-year line of improvements starting from $K^{14/15}$, computationally efficient algorithms without a simulator or exploratory assumptions still only attain $\\tilde O(K^{3/4})$ regret. A common challenge in both settings is to estimate the Q-function of a changing policy: past data comes from earlier policies that differ from (and are correlated with) the current one, while freezing the policy for future episodes worsens the dependency on $K$. We instead propose a general technique called introduced feedback delays: we estimate the Q-function of the policy using future episodes while still updating the policy, as if the feedback were delayed. While feedback delays are usually an obstacle imposed by the environment, here they are introduced voluntarily and, perhaps surprisingly, come at no extra cost. We develop Geometric Resampling with Rejection (GRR) to account for policy changes during the delay, and threshold the visitation probabilities to avoid long delays on rarely visited state-action pairs. With this unified technique, our policy optimization algorithms achieve, with high probability, $\\tilde O(\\text{poly}(H)\\sqrt{SAK})$ regret in tabular AMDPs, resolving the open problem; and $\\tilde O(\\text{poly}(d,H)\\sqrt K)$ regret in linear AMDPs, giving the first efficient algorithm optimal in $K$. They also naturally extend to delayed AMDPs with a total feedback delay $\\mathfrak D$, yielding the first bounds optimal in $S$ and $\\mathfrak D$ (tabular) and in $K$ and $\\mathfrak D$ (linear).","authors":["Aarush Kulkarni","Khang Nguyen","William Chang","Yan Dai"],"cats":["cs.LG","cs.AI","stat.ML"],"primary":"cs.LG","published":"2026-11-12T18:10:46Z","topic":1},{"id":"2611.17212","title":"Theoretical and Empirical Analysis of Sparse Convolution","summary":"Deep neural networks for image processing are increasingly deployed on resource-constrained edge devices such as drones, medical systems, and smartphones, where limited computational resources and memory bandwidth pose significant challenges for efficient inference. Pruning during training often produces highly sparse convolutional kernels, offering the potential to significantly reduce computation and memory requirements. However, exploiting this sparsity efficiently at runtime remains challenging due to irregular memory access patterns and workload imbalance across processing elements. In this work, we investigate the standard convolution with sparse weights and analyze its performance characteristics through roofline analysis, showing that a 3 x 3 sparse convolution achieves an arithmetic intensity of approximately 10 FLOP/byte compared to roughly 96 FLOP/byte for its dense counterpart. We propose techniques for an efficient sparse convolution at high sparsities. Preliminary results demonstrate that our approach achieves upto 1.65x speedup as compared to dense convolution on NVIDIA A100 and upto 1.52x on NVIDIA Jetson Nano (Tegra X1) , motivating further exploration of direct convolution as a practical path to efficient sparse inference on modern accelerators/edge devices.","authors":["Keshvi Tuteja","Yu-Hsiang Tsai","Gregor Olenik","Markus Götz","Achim Streit","Hartwig Anzt","Charlotte Debus"],"cats":["cs.PF","cs.DC","cs.LG"],"primary":"cs.PF","published":"2026-11-12T18:10:06Z","topic":4},{"id":"2611.17199","title":"MALNISI-Infused Graph Construction and Weighted Contrastive Loss based GAT for Unsupervised Representation Learning in HSI","summary":"Hyperspectral images (HSI) possess discriminating spectral characteristics with respect to land utilization due to high-dimensional information. Recently, unsupervised representation learning using graph neural networks has unveiled these utilization-specific complex spectral relations. This article has three main contributions. First, it proposes an index, the Mean Absolute Local Neighborhood Inverse Spectral Index (MALNISI), which measures the similarity between two pixels based on their neighborhood information. The index is used to construct a weighted adjacency matrix, giving the HSI a graph structure. Second, it proposes an upgraded Graph Attention Network that learns compact latent representations of hyperspectral data from the constructed graph. In this network, an enhanced multi-head attention mechanism incorporates the MALNISI weights as an auxiliary feature to emphasize learned attention between neighboring pixels. To alleviate over-smoothing and preserve fine-grained spectral details, the architecture integrates residual connections, skip projections, and feature normalization, which together enhance representation capability, stabilize learning, and accelerate convergence. Third, it proposes a weighted contrastive loss function that incorporates neighborhood-aware MALNISI between connected pixels, encouraging strongly connected pixels in the high-dimensional space to remain close in the latent space, and weakly connected pixels to stay apart. The resulting discriminative low-dimensional embeddings are then clustered using spectral clustering. Experimental analysis on six benchmark HSI datasets demonstrates the efficiency and necessity of the three aforementioned contributions, and the comparative study highlights superior performance over fourteen state-of-the-art approaches.","authors":["Soumi Sarkar","Subhadip Boral","Sarbani Palit","Ashish Ghosh"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T18:02:53Z","topic":3},{"id":"2611.17196","title":"Physically Interpretable Basin Selection in Discrete-Variable Origami Design","summary":"Automatic differentiation has enabled rapid progress on continuous-variable mechanical design problems, but gradient-based optimization still struggles with discrete variables. The standard approach is to optimize using a continuous relaxation of the choices, then commit to the nearest discrete design at the end. Why this procedure selects unfavorable basins is poorly understood. Here, we introduce a practical, interpretable design problem inspired by origami, in which automatic differentiation is used to design shapes with a target functional behavior. Each origami sheet has a fixed geometry, and a bilevel optimization framework tunes which one of four discrete hinge types to place at each fold. Baseline gradient-based design settles into basins with a large discretization gap: relaxed optimization solves 97% of runs, but the final, discrete solutions succeed only half the time. We find that the gap can be relieved with a surprisingly simple modification. An initialization prior uniformly biased toward the stiff, flat hinge option boosts the final solve fraction from 49% to 69%. Applying the prior adds no additional compute and leaves the optimization process unchanged. Compared to the baseline, sheets produced with the prior have greater geometric compatibility and mechanical stiffness. Directly penalizing frustration and floppiness in the outer loss function produces an even greater improvement than the prior does, raising the solve fraction to 78% and indicating that these properties help distinguish the unfavorable basins. Our results demonstrate that loss landscapes for mechanical design can have physically meaningful basins that are not robust to discretization and cause solves to fail. Targeting these basins' physical signatures, even with just an initialization bias, helps avoid them.","authors":["Lucy Liu","Leon M. Kamp","Katia Bertoldi","L. Mahadevan"],"cats":["cs.LG","cond-mat.mtrl-sci","cond-mat.soft"],"primary":"cs.LG","published":"2026-11-12T17:59:43Z","topic":1},{"id":"2611.17191","title":"Maximizing Capacity Utilization of LoRA Adapters for Mitigating the Learning-Forgetting Trade-off","summary":"Low-Rank Adaptation (LoRA) often underperforms full fine-tuning, especially in heterogeneous and continual-learning settings in which substantial forgetting occurs. In general, it suffers from an undesirable learning-forgetting trade-off across diverse fine-tuning settings. We identify a potential contributor: LoRA adapters, especially high-rank ones, frequently fail to realize the available capacity afforded by the configured rank. Specifically, we find that standard LoRA updates have effective rank far below their configured rank, limiting how broadly the update energy is distributed. Therefore, we propose training mechanisms that optimize the original objective while maximizing adapter effective rank. Our main method adds a simple, stable, differentiable, and SVD-free Rényi-2 effective-rank regularizer to the language-modeling loss. We also connect capacity maximization to norm-bounded updates. In mixed-domain rank sweeps on three models evaluated with multiple random seeds, a penalty weight of $λ=0.01$ lowers mean Wiki perplexity at every tested rank of at least $16$ relative to unpenalized LoRA. Mean training perplexity improves or increases by at most $0.02$ in all but one model-rank comparison. Single-task and scheduling experiments further show that the balance between learning and retention depends on the configuration. We also observe that several LoRA variants in the literature also suffer from the identified capacity issue, and our approach provides a complementary axis of improvement to LoRA-based parameter-efficient fine-tuning. These results suggest that effective rank is not only a post-hoc diagnostic of adapter quality, but a trainable objective for improving the learning-forgetting frontier. Our code is available at https://github.com/tomohiro-sawada/lora-capacity-utilization","authors":["Tomohiro Sawada","Kartik Goyal"],"cats":["cs.LG","cs.CL"],"primary":"cs.LG","published":"2026-11-12T17:55:33Z","topic":4},{"id":"2611.17186","title":"Cortex-FL: Agent-Mediated Collaboration for Federated Learning","summary":"Federated learning (FL) enables independent data owners to train models collaboratively without centralizing raw data. However, identifying collaborators, preparing training tasks, and coordinating execution and incentives hinder its practical adoption by organizations without specialized expertise. We present Cortex-FL, an agent-native FL system that enables users to express learning objectives in natural language and delegate the collaboration required to train a model. Agents translate these objectives into shared tasks, discover collaborators, and negotiate participation agreements, connecting collaboration formation with distributed training and incentive management. The architecture separates agent coordination from local execution authority, allowing owners to delegate coordination while retaining control over their data, computation, and financial commitments. We demonstrate the complete workflow on a four-node deployment, in which a model requested in natural language proceeds through collaboration formation, contract-governed training, verified model delivery, and x402 settlement on a blockchain testnet.","authors":["Rui Sun","Jinfeng Peng","Zehua Cheng","Jiahao Sun","Zhipeng Wang"],"cats":["cs.AI","cs.MA"],"primary":"cs.AI","published":"2026-11-12T17:52:51Z","topic":0},{"id":"2611.17169","title":"Parsed, Plausible, Wrong: Output Format Controls Silent Errors in Small-Model Semantic Parsing","summary":"A small fine-tuned parser that emits a well-formed but wrong tree fails silently: nothing downstream can tell. We ask whether the output format, a free design choice, controls how often this happens. On the PIZZA ordering benchmark we fine-tune two sub-billion models (Qwen3.5-0.8B and LFM2.5-350M) with LoRA on six lossless encodings of the same target tree, from 44.3 down to 10.9 tokens per order, including two matched-length pairs that vary surface form at fixed length. Across 300 runs (5 training pool sizes, 5 seeds) with a hierarchical bootstrap and Holm correction fixed before the results were read, shortening alone never helps, but writing the target in ordinary words does. The word encoding is the most accurate format on both models (87.5% vs. 85.8% exact match for the native s-expression on Qwen) while emitting 4.1x fewer tokens, and it cuts the silent error rate by 2.3x on Qwen and 1.8x on LFM2, separating at every pool size. Error analysis locates much of the gain in negation. All runs also learn to refuse off-menu input (held out: 95.9-99.0% handled) and to apply in-utterance self-corrections. Strict refusal has a measured price: 2.1-10.1% of real, mostly polite, orders are refused.","authors":["Umar Ahmed","Maxim Dunaev","Mohammad Nour Al Awad","Sergey Ivanov","Ivan Khodnenko"],"cats":["cs.CL","cs.LG"],"primary":"cs.CL","published":"2026-11-12T17:43:44Z","topic":4},{"id":"2611.17167","title":"Actor, Judge, or Teacher? Understanding Decision Models in Long-Horizon Agents","summary":"Decision models answer a question with a typed output, such as a choice among options with a probability for each, so one model can choose an agent's actions, evaluate actions already taken, or score every action to teach a policy with reinforcement learning. Prior work has studied them mainly as actors, judging their usefulness by how well they act. We ask whether a decision model that acts poorly can still teach well. We compare three decision models and three general LLMs as actors and judges on ALFWorld, and use JEV as a teacher on ALFWorld and WebShop. As actors, all six models are pulled away from a subgoal that must come first by an easier one. As judges that see the whole trajectory, the same models credit the steps toward that subgoal, and asking whether each action helped, rather than whether the task will succeed, separates useful from useless actions. As a teacher, JEV raises a Qwen2.5-1.5B policy over GRPO from 72.8% to 82.9% success on ALFWorld and from 56.8% to 71.4% on WebShop. These results suggest that a poor actor can still teach well, because judging an action after the episode ends reveals the consequences that acting must anticipate. Code is available at https://github.com/ZJU-REAL/decision-agents.","authors":["Haochen Liu","Qianhao Zhu","Zixuan Wang","Dingming Li","Tongbo Chen","Peng Qiu","Yingqi Qu","Weiming Lu","Jun Xiao","Yongliang Shen"],"cats":["cs.AI","cs.CL"],"primary":"cs.AI","published":"2026-11-12T17:42:46Z","topic":0},{"id":"2611.17162","title":"Symmetry in Time: A Representation Theory of Linear Recurrent Memory","summary":"State space models (SSMs) are an emerging alternative to Transformers. SSMs offer linear-time inference and parallel training, but most implementations make the parallel scan efficient by making the recurrence diagonal: each channel of the hidden state is multiplied by an input-dependent real or complex number. These numbers commute, so the state records which inputs occurred but not their order. We formalize this with representation theory, which explains and fixes the state-tracking failures previously reported for SSMs. The state update defines a linear representation of the monoid of input sequences, and we show that an SSM tracks a group at every sequence length exactly when that group is a quotient of the group generated by its transition matrices. Two consequences follow. Diagonal real- and complex-valued SSMs can track only commutative groups, while the transition matrices of any SSM that can track a group must split into independent blocks over the real, complex, and quaternion numbers. We then introduce HyperSSM, which adds quaternion channels while keeping linear-time inference and parallel training. Trained only on final answers, HyperSSM solves the group tasks of recent state-tracking benchmarks exactly with at most five state numbers. This includes $S_5$ with 32 generators, where previous SSMs are at chance, and HyperSSM generalizes to sequences 25 times longer than those seen in training. Because the theory says what the transitions must encode, the learned weights of the linear recurrence can be interpreted directly. We demonstrate this by recovering the calibration of a real gyroscope, the rotations of imperfect qubit gates and the geometry of protein backbones.","authors":["Anif N. Shikder","Ramit Dey","Jan Minac","Lyle E. Muller"],"cats":["cs.LG","cs.NE","math.RT"],"primary":"cs.LG","published":"2026-11-12T17:39:35Z","topic":4},{"id":"2611.17161","title":"Cited but Unused: Parametric Provenance in Retrieval-Augmented Generation","summary":"Retrieval-augmented generation systems attach citations to attribute claims to their sources. These citations frequently misstate provenance, and standard verification cannot detect this. We measure provenance by counterfactual value substitution: a fact in the retrieved passages is replaced with a false variant and the model re-queried. We find that across five sizes of one model family and a second comparator family, 1.2% to 15.7% of citable claims derive from parametric memory rather than the retrieved context. At the most conservative substitution distance these figures are 4.6% and 8.5% on the two models we tested at it, with the rate rising with model size within a family. We further find that natural language inference verification, the standard approach, gives results indistinguishable from chance across models, even when restricted to the cited passage. We determine that the cause is structural: a model answering from memory selects a citation that agrees with its output. Comparing ten attribution methods on identical claims, the contrast between the model's distribution with and without context best separates provenance. Acting on that signal does not improve provenance honesty at these base rates. The resulting detector is hence still not accurate enough to act on: it measures what the model knew rather than what it used, and separating those remains the open problem.","authors":["Tanzeel Abbas","Tehseen Zia","David Windridge"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T17:39:14Z","topic":6},{"id":"2611.17152","title":"AeroZoo: A Benchmark of Well-Posed Flow Problems for Foundation Models of Aerodynamics","summary":"Neural surrogates for computational fluid dynamics (CFD) are trained and ranked on datasets that each vary one design family in one flow domain. Such datasets cannot test the least that an aerodynamic foundation model must do: predict CFD-quality incompressible flow in seconds for new shapes in new enclosures. On DrivAerML, larger models and longer training barely lower Transolver's error; doubling the training cases cuts it by about 12% up to 400 cases, exhausting the dataset. We introduce AeroZoo, a benchmark and training corpus of 3,815 time-mean solutions to the unsteady Reynolds-averaged Navier--Stokes (RANS) equations with a transition model, across seven geometry families, eight enclosure configurations and Reynolds numbers from $10^3$ to $10^8$. Each sample poses a well-posed PDE problem in the form a solver receives it: the boundary meshes grouped by condition type, the prescribed values and the fluid properties. The 2,816 samples with a stable time mean form the canonical benchmark: two tasks, engineering metrics and 22 frozen splits, disjoint in geometry, that hold out geometry families, enclosures, boundary-condition layouts and Reynolds ranges. On the random split, boundary-condition tokens cut GeoTransolver's median velocity error from 76.9% to 8.4% of the inlet speed in single-seed runs (uniform inlet flow: 11.0%). ISLA, a transformer for boundary-driven PDEs, reaches 5.3% over four seeds, about 4 times the change from a finer mesh. On the same test samples, a held-out geometry family barely changes ISLA's error relative to uniform flow, and a held-out enclosure or boundary-condition layout raises it 2.4 to 9.2 times. The data and the scoring code will be released.","authors":["Peter Sharpe","Carmelo Gonzales","Corey Adams","Mehdi Ataei","Sanjay Choudhry"],"cats":["cs.LG","cs.CE","physics.flu-dyn"],"primary":"cs.LG","published":"2026-11-12T17:36:09Z","topic":6},{"id":"2611.17145","title":"Enhancing Preference Detection using Eye Movement Cascade and Multilayer Combinatorial Fusion","summary":"Understanding how visual attention influences human preference formation is a fundamental problem in consumer behavior and cognitive science. In this work, we investigate how eye movement patterns can be used to predict preference decisions when subjects compare two alternatives. To address this problem, we analyze an eye-tracking dataset, where subjects view pairs of face images and their gaze behavior is recorded throughout the decision process. We model this problem using multiple machine learning classifiers and apply the Multi-layer Combinatorial Fusion (MCF) framework to iteratively combine diverse models. The MCF framework expands models to a large pool of model fusions and then reduces them based on cognitive diversity, repeating this process across layers until diversity diminishes. Experimental results show that MCF significantly improves performance over individual models. The best single model achieves a Pre@k of 0.90, which is improved to 0.9143 after the first layer of MCF, and further increased to 0.9286 in the second layer. This progression demonstrates that the second layer not only surpasses the performance of the first layer but also outperforms the best individual model, highlighting the effectiveness of leveraging diversity through multi-layer fusion.","authors":["Yuanhong Wu","Mary Tsaryk","Sam Watson","Tahsin Ahmed","Christina Schweikert","D. Frank Hsu"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T17:31:50Z","topic":3},{"id":"2611.17143","title":"A Convex-Analytic Reformulation for Poisoning Attacks on Logistic Regression","summary":"We study data poisoning attacks on logistic regression with $\\ell_2$-regularization, where an attacker chooses the features of a fixed number of injected observations with prescribed labels to maximize the logistic loss on the clean training data. Strong convexity of the defender's problem yields an exact reformulation through its first-order optimality conditions. Using convex conjugacy of the softplus function, we replace the sigmoid relations by a single convex optimality-gap inequality and express the attacker's objective as a concave function. A change of variables makes the stationarity condition affine, leaving bounded bilinear equalities as the only nonconvex constraints. We derive finite bounds on all variables involved in these equalities, including a global bound on the coefficient norm and tighter bounds obtained by preprocessing based on convex optimization. Experiments on eight binary classification datasets evaluate attack effectiveness and compare the two reformulations when solved with \\textsf{SCIP}. Attacks obtained by local optimization degrade predictive performance on most datasets, while the convex-analytic formulation produces substantially stronger reported dual bounds within a fixed time limit.","authors":["Carina Moreira Costa","Maria Eduarda Pinheiro","Martin Schmidt"],"cats":["cs.LG","math.OC"],"primary":"cs.LG","published":"2026-11-12T17:30:34Z","topic":7},{"id":"2611.17141","title":"ISLA: A Coordinate-Free, Measure-Weighted Transformer for Boundary-Driven PDEs","summary":"A foundational principle of physics is that all coordinate systems are equally valid: the behavior of a PDE-governed system does not depend on the origin or orientation of the frame, nor on the units or mesh used to write it down, and neither should a surrogate model. Yet the neural surrogates with the lowest errors on today's aerodynamics benchmarks lack these invariances and fail under simple transformations of the same problem: for models trained on PDEs of cars in one pose (i.e., translation and rotation) and from one mesher, rotating the full PDE raises the surface-pressure error from 4% to more than 100%, and re-sampling its boundary raises it to about 30%. We introduce ISLA (Invariant Slice Attention), a slice-attention transformer for steady, boundary-driven PDEs in which no learned operation receives a coordinate and every sum over the boundary is a quadrature. Positions enter only as distances and relative angles to learned slice anchors, lengths are measured in a unit computed from the boundary, and interior points query the boundary through a tree-accelerated learned kernel without influencing one another. With every boundary cell a token and an exact boundary read, ISLA is equivariant to rigid motions and reflections, invariant to the units of its declared data, and consistent under mesh refinement, while at every setting each query is independent of all other queries. On AeroZoo, 2,816 flows that vary in shape, size, speed, and boundary conditions, its mean reference-scaled velocity and pressure errors are at least 3.1 and 3.3 times lower than any baseline's. On AirfRANS its errors are second only to those of a boundary-integral model (GLOBE) and at least 1.6 times lower than those of Transolver and GeoTransolver.","authors":["Peter Sharpe"],"cats":["cs.LG","cs.CE","math-ph","physics.flu-dyn"],"primary":"cs.LG","published":"2026-11-12T17:30:03Z","topic":6},{"id":"2611.17134","title":"Support-Constrained Offline Policy Improvement for Pediatric ECMO from Limited Clinical Data","summary":"Managing children on ECMO (Extracorporeal Membrane Oxygenation) requires balancing short-term physiological stability with the long-term goal of preventing neurological injury. Standard reinforcement learning can propose risky, out-of-distribution actions, while imitation learning is limited to reproducing clinician behavior. We introduce Support-Constrained Fitted Q-Iteration (FQI-SC), an offline reinforcement learning framework that cautiously improves upon expert behavior by restricting optimization to empirically supported state-action pairs and defaulting to the clinician baseline in low-data regions. On 78 pediatric trajectories totaling over 6,000 hours of clinical data, we evaluate our framework using Fitted Q-Evaluation (FQE) and a behavioral disagreement. Additionally, we formalize the performance-safety trade-off by introducing a unified Balance Score ($S(π)$). Our empirical evaluation shows that support-constrained policy improves the Balance Score by 38\\% relative to FQI, improves estimated value by 1.28 FQE units over imitation learning, and reduces disagreement with clinician actions by 71\\% relative to FQI.","authors":["Fateme Golivand Darvishvand","Saurabh Mathur","Nikhilesh Prabhakar","Michael A. Skinner","Ameet Soni","Neel Shah","Ethan L Sanford","Lakshmi Raman","Sriraam Natarajan"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T17:26:01Z","topic":1},{"id":"2611.17129","title":"Paper-Faithful Federated Learning: Scientific Checks for Practitioner and Agent Development","summary":"Federated learning experiments can produce reasonable-looking results even when the experiment implementation differs from the intended experimental protocol. Small errors in configuration, data partitioning, aggregation, randomness, evaluation, or checkpoint recovery may therefore affect a comparison without causing the program to fail. We present FedBrew, a research codebase for controlled and reproducible federated learning experiments. Its focus is controlled simulation with checks of specified conditions of the intended experimental protocol. FedBrew uses configuration checks, identifiable client data partitions, checkpoint recovery checks, self-describing run records, and development-time checks of scientific conditions. These checks assess, for example, whether client partitions are disjoint and exhaustive and aggregation weights match the prescribed rule; numerical and recovery checks examine non-finite client model states and interrupted runs. The same structure also provides clear extension points and automated checks for new algorithms, datasets, models, and tasks, whether they are added by researchers or AI coding agents. FedBrew therefore supports reproducible and inspectable FL experiments through checks of specified software and scientific conditions and the run records associated with reported results.","authors":["Ali Dadras","Alp Yurtsever"],"cats":["cs.LG","cs.SE"],"primary":"cs.LG","published":"2026-11-12T17:22:17Z","topic":0},{"id":"2611.17128","title":"Parameter-Efficient Conciseness Adaptation: Differentiable Length Control for Concise Clinical Reasoning in Smaller Language Models","summary":"Smaller language models (SLMs) make it feasible to deploy clinical natural language processing where large proprietary models are too costly, and parameter-efficient fine-tuning allows them to be specialised on commodity hardware. One obstacle persists after training: on medical tasks these models answer at far greater length than the clinical question requires. In clinical decision support and point-of-care use, where time is short, such verbosity is hard to act on even when the underlying reasoning is sound. Existing length-control methods depend on preference optimisation, reward models, or proprietary teachers that undercut the accessibility making SLMs attractive. We introduce Parameter-Efficient Conciseness Adaptation (PECA), which adds a single differentiable conciseness term to the QLoRA supervised fine-tuning objective. Because the token count is discrete and non-differentiable, PECA instead raises the end-of-sequence probability once generation exceeds a token budget, a smooth signal optimised during ordinary training, with no preference data, no auxiliary model, and a single commodity GPU. On a 1.7-billion-parameter model fine-tuned on the medical-o1-reasoning-SFT corpus, PECA reduces mean response length by 72.8% while improving multiple-choice diagnostic accuracy from 40.8% to 44.0%. A sweep over the conciseness strength traces an interpretable accuracy-to-length frontier for balancing brevity and diagnostic completeness.","authors":["Tanzeel Abbas","Ahmad Kamran Malik","Nafees Qamar","Sana Liaqat"],"cats":["cs.CL","cs.AI"],"primary":"cs.CL","published":"2026-11-12T17:21:12Z","topic":4},{"id":"2611.17126","title":"STeP-C: Real-Time Neural Compression for Distributed Fiber-Optic Sensing at Scale","summary":"Distributed fiber-optic sensing (DFOS) technologies such as Distributed Acoustic Sensing (DAS) generate continuous, high-resolution spatio-temporal data at rates exceeding hundreds of megabytes per second, creating severe challenges for real-time storage and transmission. While neural compression methods can achieve strong rate-distortion performance on sensing data, prior approaches either require per-segment retraining incompatible with streaming, or have only been validated offline. None have been demonstrated end-to-end on a continuously streaming DAS acquisition system at scale. We present STeP-C (Spatio-Temporal Predictive Compression), a neural compression framework that decouples spatial encoding, temporal prediction, and entropy-optimized residual coding, trained end-to-end with a rate-distortion objective motivated by conditional entropy reduction. Critically, we identify entropy coding as the dominant bottleneck for streaming deployment. We address it through GPU-side symbol computation, cached probability tables, and parallel GIL-free arithmetic coding, together with FP16 inference and batched entropy coding, enabling a full real-time pipeline built around a GPU ring buffer. We validate this pipeline on a live DAS testbed, achieving a 98.85% file size reduction at 31.04 dB full-band PSNR and an average throughput of 535 MB/s, exceeding the 324 MB/s requirement for 1 kHz, 81 km DAS streams. We further show that STeP-C substantially outperforms conventional codecs at practical streaming bitrates, can be applied to unseen channel ranges within the same acquisition system without retraining, and largely maintains downstream detector consistency, with 96.62% high-confidence class agreement and 100% Jaccard similarity on detected event types.","authors":["Tingfeng Li","Ming-Fang Huang","Shaobo Han","Yue Tian"],"cats":["cs.CV","cs.LG"],"primary":"cs.CV","published":"2026-11-12T17:20:49Z","topic":4},{"id":"2611.17121","title":"Using an energy pixels approach to estimate the size a 2nd Generation DEL must be to train a high-resolution binding model","summary":"DNA-encoded libraries (DELs) are commonly used to find small-molecule binders to a target protein. But the hard part is no longer finding a hit. It is building a machine-learning model good enough to guide the medicinal chemistry that follows, one that can resolve the shift in affinity when a chemist changes a single atom. Such a model needs two things from its training data: dense sampling of the small patch of chemical space around the hit, and an adequate number of genuinely tight binders within it. A diverse naive DEL gives neither. It samples an enormous space, so any one neighborhood is covered far too thinly, and the tightest binders are rare. A model trained on that naive DEL data must extrapolate to binders tighter than any it has seen, and that is where models can fail. One way to get training data with both properties is a second-generation DEL built around the hit. We ask how complex that library must be. We treat binding as a sum of small, independent interactions, called pixels on the target and antipixels on the ligand. From the spread of affinities measured in the naive library, with no structural information and nothing yet synthesized, we estimate the complexity such a library must have to supply that dense sampling and those tight binders. The method does two useful things. It turns the number of tight binders a model needs into the number of compounds a library must contain. And it estimates the expected yield of a randomly designed second-generation DEL, so the gain from designing the library with machine learning or active learning can be measured against a baseline, without the expense of a control library as large as the experiment itself.","authors":["Boris P. Belotserkovskii","R. Edward Watts"],"cats":["cs.LG","q-bio.BM"],"primary":"cs.LG","published":"2026-11-12T17:13:57Z","topic":7},{"id":"2611.17116","title":"Bellman Contraction under MMD: A Unified Framework","summary":"We introduce a mathematical framework for studying contraction of the distributional Bellman operator under Maximum Mean Discrepancy (MMD). We show that contraction reduces to conditional positive definiteness of an associated kernel constructed from the base kernel, which we call the Bellman difference kernel. Focusing on radial kernels, we use Schoenberg's theorem to connect this condition to Bernstein functions. This gives contraction criteria that hold in any dimension and can be verified analytically or tested numerically. Our framework unifies known results for the Gaussian RBF, power-distance, and multiquadric kernels. A numerical search identifies 11 new kernels that induce contraction. We also propose chaining, a general construction that produces kernels inducing contraction with any prescribed factor. To reconcile theory with practice, we show that contraction on a set only requires conditional positive definiteness on that set. This explains why the RBF kernel can remain a valid choice when rewards are bounded, despite its lack of global contraction. Finally, we identify a fundamental tradeoff between contraction rate and discriminative power, where kernels that contract faster tend to discriminate less between distributions. Code is available at https://github.com/BaptisteDebes/BellmanContractionMMD","authors":["Baptiste Debes","Tinne Tuytelaars"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T17:09:17Z","topic":2},{"id":"2611.17115","title":"CapLite: Prompt-Injection Defense for Small Language Model Agents without Model-Written Security Artifacts","summary":"Enforcement defenses against prompt injection, such as CaMeL, Progent, IPIGuard and DRIFT, ask the agent's own model to write a security artifact (a program, a privilege policy, a dependency graph or a constraint list) that code then enforces. We ask whether small language model (SLM) agents can be trusted to write these artifacts. Across 15 open-weight models, 13 of them between 1.7B and 12B parameters, they cannot: the policies, graphs and constraints that the agent's model writes admit attacks on 14 (Progent), 13 (IPIGuard) and 11 (DRIFT) of the models, and CaMeL's programs fail to run in up to 68% of episodes. We present CapLite, which derives its policy from the tool schemas, commits the write calls before any tool result is read, and admits a write only if every argument that decides whom or what it affects is stated in the user's request. This guarantees write integrity: injected text cannot add a write, and no write reaches a recipient, amount, object or destination that the user did not state. No attack succeeds in 13,935 attacked AgentDojo episodes. CapLite reaches utility under attack comparable to CaMeL's (29.0% vs. 27.2%) with fewer tokens than every other enforcement defense and than the undefended agent: 17.1k per episode, 18% of CaMeL's, and the highest utility per token of all enforcement defenses.","authors":["Zimo Xu","Wenzhuo Liu","Hideki Nakayama"],"cats":["cs.AI","cs.CL","cs.CR"],"primary":"cs.AI","published":"2026-11-12T17:07:22Z","topic":0},{"id":"2611.17114","title":"Designing Institutions for Autonomous AI","summary":"As autonomous AI agents increasingly interact in persistent populations, their collective capabilities and alignment depend not only on individual models, but also on the institutions governing their interactions. Yet AI research has largely focused on individual agents. In this Perspective, we argue that institutional design must become a distinct focus of AI research. Drawing on institutional political economy, social epistemology and organizational theory, we develop seven principles for designing AI institutions that sustain cooperation, enable reliable collective judgments and remain accountable to legitimate human purposes. These address reciprocal contributions, accountable authority, functional differentiation, epistemic pluralism, criticism and correction, human direction, and institutional adaptation. We illustrate their application through machine science, where autonomous research challenges institutions developed around human capabilities and constraints. We then propose an experimental research programme in which matched populations of AI agents operate under alternative institutional arrangements, and describe MoltJournal, an API-first journal for autonomous research agents, as an illustration of how such institutions can be made programmable and observable. Together, these contributions establish institutional design as a research agenda concerned not only with aligning autonomous agents, but with understanding how institutions can transform them into capable, reliable and accountable collectives.","authors":["Petter Törnberg"],"cats":["cs.SI","cs.AI"],"primary":"cs.SI","published":"2026-11-12T17:06:54Z","topic":0},{"id":"2611.17110","title":"SALA: Semantically Aligned Latent Actions Enable Dual-Axis Scaling of Embodied Foundation Models","summary":"Embodied foundation models hinge on diverse robot interaction data, which are costly to collect. Egocentric human videos offer a complementary source of experience, and latent action models (LAMs) provide a popular means of leveraging these videos for policy supervision. Compared with world-action models (WAMs), LAMs can offer greater training efficiency and potentially lower inference costs by avoiding explicit pixel prediction during policy learning and execution. However, most existing LAMs learn latent actions solely through video reconstruction, which can cause their representations to cluster by visual appearance rather than by the underlying actions. As a result, they struggle to align human actions with their robot counterparts. To bridge the gap, we introduce SALA, which emphasizes establishing the semantic correspondence for latent actions. Without requiring paired data, we use the language descriptions of unpaired human and robot clips to estimate semantic similarities and supervise cross-embodiment correspondence. Specifically, we translate these similarities into soft neighborhood structures that guide human latent actions toward semantically related robot latents, while robot action reconstruction grounds the latent space in control. The resulting shared latent action space provides a unified supervision signal for training an embodied foundation model on both human and robot videos. Notably, latent action supervision enables scaling along two axes: (1) pre-training on increasing amounts of general human video consistently reduces robot action prediction error on downstream tasks, and (2) post-training on increasing amounts of task-specific human video substantially improves real-world robot task success rates.","authors":["Jiachun Jin","Zetong Zhou","Haonan Wang","Xiao Yang","Jun Zhu","Zhijie Deng"],"cats":["cs.RO","cs.AI","cs.LG"],"primary":"cs.RO","published":"2026-11-12T17:04:50Z","topic":2},{"id":"2611.17109","title":"Near-Optimal Regret for Oracle-Efficient Adversarial Contextual Bandits","summary":"We study contextual bandits with $K$ actions, a class of $N$ policies, and i.i.d.\\ contexts from a distribution that the learner can sample. For any sequence of loss functions fixed before play, we give an oracle-efficient algorithm with expected regret $O(\\sqrt{KT\\log(2N)}\\,\\log(2KT))$ against the best policy in hindsight on the realized contexts. This improves the $T^{2/3}$ dependence on the horizon in prior oracle-efficient bounds and achieves the optimal regret rate in this setting up to logarithmic factors. Our algorithm perturbs policy scores using a Poisson sample of auxiliary contexts. At each sampled context, the perturbation assigned to an action is a Laplace noise draw multiplied by a scale specific to that context and action. We show that choosing these scales to minimize a convex objective determined by past observations yields the stated regret bound. We further approximate the minimizing scales by projected stochastic gradient descent using only an exact offline cost-sensitive optimization oracle that returns the optimum value. Since the table of scales can be large, the algorithm represents the SGD iterates implicitly by storing the random draws that determine each update and evaluates the scales only at queried contexts. Both the oracle complexity and the remaining resource bounds are polynomial in $T$, $K$, and $\\log(2N)$.","authors":["Kiarash Banihashem","MohammadTaghi Hajiaghayi"],"cats":["cs.LG","cs.DS"],"primary":"cs.LG","published":"2026-11-12T17:04:36Z","topic":1},{"id":"2611.17108","title":"LazyWeights: Decode-on-Demand Edge LLMs","summary":"Memory decides which LLMs an edge device runs, and how fast: capacity on a $4$ GB Jetson Nano, bandwidth on an $8$ GB Jetson Orin Nano Super. Deployed formats (GPTQ, AWQ, $\\texttt{llama.cpp}$'s K- and I-quants) spend equal bits on every integer code of a matrix, although the codes are far from uniform. $\\texttt{LazyWeights}$ keeps these formats but tunes GPTQ so the codes compress, Huffman-codes them losslessly in interleaved lanes, and decodes them on demand in a small unit between DRAM and L2, so GPU kernels run unchanged and DRAM moves only coded bytes. At $1\\%$ and $3.5\\%$ perplexity budgets, coded Phi-3-mini, Mistral-7B and SmolLM-1.7B are $5.7-31.4\\%$ smaller than the best $\\texttt{llama.cpp}$ format. A bit-exact $0.21$ mm$^2$ ASIC decoder unit ($1$ GHz, $7$ nm) cuts Orin latency and energy per token by up to $36\\%$ and $38\\%$ ($54\\%$ and $44\\%$ with prompt-lookup speculation). On the Nano, $\\texttt{LazyWeights}$ needs $58-64\\%$ less time and $53-57\\%$ less energy per token than CPU $\\texttt{llama.cpp}$, and runs a streamed Mistral-7B 6.6$\\times$ faster at $3.89$ bits per weight. A Huffman decoder beats an rANS decoder in rate and area at equal throughput; the ASIC decoder unit runs bit-exact on a Zynq UltraScale+ FPGA (AMD ZCU102).","authors":["Arnab Sanyal","Gourav Datta","Sandeep P. Chinchali"],"cats":["cs.LG","cs.AR","cs.CL","cs.IT"],"primary":"cs.LG","published":"2026-11-12T17:03:40Z","topic":6},{"id":"2611.17107","title":"Disentangling Spurious Correlations: How IRM Handles Diverse Shortcuts in Text Classification","summary":"In NLP, models trained via Empirical Risk Minimization (ERM) often exploit spurious correlations, such as dialectal markers, at the expense of robust linguistic features, thereby degrading their out-of-distribution (OOD) generalization. Invariant Risk Minimization (IRM) aims to suppress such shortcuts by identifying predictive relationships that remain stable across environments, but its practical utility for NLP remains debated. We use a causality-driven framework to evaluate IRM under three shortcut-generating mechanisms: label-generated features, confounding, and selection bias. Our results show that IRM successfully rejects shortcuts across the evaluated causal and anti-causal structures, including under limited environmental variation and increasing label noise. We then extend our analysis beyond controlled synthetic artifacts to two less constrained settings: gender-profession associations in Bias in Bios and semantically loaded actor-name shortcuts in ActorCorr. These experiments show that invariant learning can mitigate more realistic textual shortcuts, while semantically meaningful features require careful calibration of the invariance penalty.","authors":["Paul Grunenwald","Benoît Crabbé","Guillaume Wisniewski","Pirmin Lemberger"],"cats":["cs.CL","stat.ML"],"primary":"cs.CL","published":"2026-11-12T17:03:05Z","topic":6},{"id":"2611.17105","title":"Predicting Ethanol Drinking Status from Nonhuman Primate Liver Histology: A Proof of Concept for Parameter-efficient Fine-tuning in Few-Shot Vision-Language Modeling","summary":"Combining vision-language models (VLMs) with parameter-efficient fine-tuning (PEFT) offers a promising approach to few-shot whole-slide image (WSI) classification in computational pathology. This study evaluates this approach for predicting ethanol drinking status from nonhuman primate liver histology. Adapters are evaluated against non-PEFT configurations on a subset of the Monkey Alcohol Tissue Research Resource (MATRR). All evaluated PEFT configurations achieve higher mean balanced accuracy than the non-PEFT baselines. For example, in the four-shot setting, attention-based multiple instance learning (ABMIL) alone achieves 43.5% mean balanced accuracy, compared with 78.1% after adding an adapter. Qualitative attention analysis shows that textual pathway adaptations produce localized attention heatmaps, while visual pathway adaptations produce more diffuse ones. A pathologist's review identified similar histological features across adaptations.","authors":["Amirmohammad Mohammadi","Rupak Khadka","Kathleen A. Grant","Mary Lauren Benton","Joshua Peeples"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T17:02:27Z","topic":3},{"id":"2611.17103","title":"Joint Null-Space Projection for Data-Free Model Merging","summary":"Merging independently fine-tuned checkpoints into a single multi-task model requires no data, no labels, and no gradient steps. In practice the task vectors of different tasks interfere, and the standard remedies--averaging, magnitude pruning, sign election--operate entry by entry, ignoring that each delta induces a low-dimensional update on the hidden features it acts upon. We recast merging as a subspace interference problem. For every two-dimensional weight tensor we take the top-r right-singular subspace of each task's delta--the input directions along which that task moves the representation most--and remove from each task vector the component lying in the union of the other tasks' subspaces. The resulting Joint Null-Space Projection (JNSP) is closed-form and order-independent, needs no forward pass, and is defined entirely by the checkpoint set. On eight CLIP image-classification tasks JNSP merges ViT-B/16 experts to 89.45% average accuracy without using a single image, 11.88 points above the strongest parameter-wise training-free baseline (TIES-Merging, 77.57%), 14.22 above Task Arithmetic, and 0.98 above WUDI's data-free consensus merge (88.47%). The gain holds across three backbones on the independently trained TALL release (+10.8, +12.6, +9.1 over TIES for ViT-B/16, ViT-B/32, ViT-L/14) and replicates there at 89.20%; controls confirm the source--a single shared subspace instead of per-task ones loses 2.9 points, and a sequential variant collapses to 69.6%. On RoBERTa-large the same projection underperforms plain task arithmetic, which we report rather than hide. Code is available at https://github.com.","authors":["Tian Xia"],"cats":["cs.CV","cs.LG"],"primary":"cs.CV","published":"2026-11-12T17:01:39Z","topic":3},{"id":"2611.17101","title":"A Compact xDAWN-Guided Spatio-Temporal Network with Filter-Bank Convolution for P300 Detection","summary":"P300-based brain-computer interfaces (BCIs) enable communication by decoding stimulus-evoked EEG responses into character selections without requiring voluntary motor output. Reliable P300 detection remains difficult because of low response amplitude and signal-to-noise ratio, while practical spellers require accurate character selection with few stimulus repetitions and low computational cost. We propose a compact deep learning architecture that combines xDAWN spatial filtering for event-related potential enhancement and dimensionality reduction, a fixed frequency-selective filter bank, and convolutional feature learning for spatio-temporal EEG decoding. Depthwise-separable convolutions and progressive average pooling reduce computational complexity while retaining discriminative temporal information. The resulting target/non-target classifier contains only 5,100 trainable parameters. We evaluated the model on two public benchmarks, BCI Competition II Dataset IIb and BCI Competition III Dataset II, and compared it with established P300 decoding methods across different numbers of stimulus repetitions. The model achieved an information transfer rate of 17.8 bits/min with two repetitions on BCI Competition III and 40.5 bits/min on BCI Competition II, demonstrating a favorable accuracy-speed trade-off. These results suggest that combining domain-informed spatial and spectral priors with compact neural networks can support efficient P300 decoding for EEG-based BCI communication.","authors":["Mohammad Mahdi Habibi Bina","Mohammad Reza Daliri"],"cats":["cs.HC","cs.AI"],"primary":"cs.HC","published":"2026-11-12T17:00:30Z","topic":4},{"id":"2611.17098","title":"Speed-to-Power Dominates Energy Costs for AI Data Centers: Technoeconomic Analysis of Powering AI in the Continental United States","summary":"The rapid growth of artificial intelligence is driving demand for new data centers at large scale and on short deployment timelines. To power them, firms connect to the commercial grid or contract generation directly, but interconnection and gas-turbine delays now incentivize on-site generation. We compare grid-connected and off-grid data center energy infrastructure (AC- and DC-coupled solar-plus-storage, islanded natural gas, and grid interconnection) across the continental U.S. Using an open-source technoeconomic framework built on empirically calibrated AI training loads, climate-driven cooling, stage-by-stage power conversion, and physics-based battery degradation, we model designs providing equivalent energy services. Our central contribution is a speed-adjusted cost of energy that values the compute unlocked by faster construction, making options with different deployment timelines directly comparable on one physically consistent basis. We find DC coupling reduces lifetime energy costs by 18% ($0.09/ kWh) on average versus AC counterparts by eliminating inverters and improving efficiency. Though rarely cheapest on unadjusted energy cost, DC-coupled microgrids' faster deployment makes them the lowest-cost modeled option in 85% of U.S. locations once foregone compute is valued, or 55% under accelerated natural gas timelines. For AI infrastructure, deployment speed rather than unit energy cost determines the economically preferred path to power.","authors":["Alex C. Newkirk","Daniel L. Gerber","Erica R. H. Fuchs","Alex Hubbard","M. Granger Morgan","Arman Shehabi","Sarah J. Smith","Constantine Samaras"],"cats":["cs.AI","eess.SY"],"primary":"cs.AI","published":"2026-11-12T16:58:53Z","topic":4},{"id":"2611.17091","title":"Optimal Single-History Tokenization Disambiguation for Provably Secure Linguistic Steganography","summary":"Provably secure linguistic steganography assumes that the receiver recovers the sender's tokens, but a text channel delivers only text, and ambiguous tokenizations break extraction. Existing fixes change the distribution, synchronize whole pools of tokens at a capacity that collapses with large vocabularies, or track several token histories. We prove that if extraction is exact and error-free across full-support language models, a single-history receiver is confined to a prefix-free candidate set at every step -- even if it reads the entire future text and the sender constrains later steps -- and must therefore give up the stop entropy, the entropy of deciding whether each prefix token ends there. Fixing these stop decisions on the prefix forest with shared randomness yields coin projection: a closed-form scheme that attains the bound, works with any distribution coder in one model call per token, preserves ordinary sampling, and refines SyncPool for every key. Across English, Chinese and Japanese, ambiguous text is rare (at most 11 in 1,000 tokens), yet the stop entropy at word and morpheme prefixes takes 7.2-9.0% of the entropy at every step; SyncPool's further loss comes from space and single-byte UTF-8 hubs that merge subtrees while carrying almost no stop uncertainty. Coin projection keeps 91-93% of the string entropy for the coder at every vocabulary size; with the full vocabulary, it turns 2.16-6.85 times as much entropy into message bits as SyncPool and recovers every completed message.","authors":["Feng Jiang"],"cats":["cs.CL","cs.CR"],"primary":"cs.CL","published":"2026-11-12T16:54:18Z","topic":6},{"id":"2611.17072","title":"Reasoning Supervision and Capability Retention in LoRA Domain Adaptation","summary":"Fine-tuning foundation models on financial data has a wide range of applications, including identifying event types from news, monitoring risk and compliance language, and answering questions over filings and earnings calls. In this setting, the main challenge is to improve task performance without degrading the generalization capabilities of the model. In this paper, we present an extensive set of ablations for fine-tuning Qwen3-8B with low-rank adaptation (LoRA). We compare three training regimes: (1) response-only LoRA, where the supervised response is the label alone; (2) single-adapter reasoning, where reasoning traces from a teacher model are included alongside the labels in the training loss; and (3) dual-adapter composite training, where separate LoRA adapters are trained on traces and labels. Response-only LoRA improves in-domain performance and leaves general capabilities near those of the frozen model. The in-domain change is largest for calibration and argmax accuracy. Under single-adapter reasoning, the in-domain metrics remain behind those of the response-only regime, while MMLU-Pro can rise slightly and the other retention benchmarks stay within ordinary variation. Dual-adapter composite training remains on the same side of this comparison and yields no further gain. Finally, we compress the teacher traces by summarizing them before training, and show that shorter summaries move calibration toward the response-only regime without reaching it.","authors":["Saleh Ashkboos","Dhruv Desai","Lavinia Ghita"],"cats":["cs.AI","cs.CL","cs.LG"],"primary":"cs.AI","published":"2026-11-12T16:45:12Z","topic":4},{"id":"2611.17068","title":"Discovering Effective Agent Memory Designs through Program Evolution","summary":"Program evolution provides a promising approach to discover agent memory designs and study the choices underlying their effectiveness. We introduce MAPE, a framework that evolves whole memory programs from minimal seeds through a standardized lifecycle interface and an iterative diagnosis-proposal loop. Experiments on LoCoMo and LongMemEval yield an absolute accuracy gain of up to 12% over manually designed baselines at 35% of the serving cost, with further evaluations supporting transfer across serving backbones and benchmarks. Empirical analyses and ablations identify and validate two recurring design patterns: retaining raw conversations as retrievable evidence preserves details omitted from extracted memories, while composing retrieved evidence according to query intent helps the answering model use it effectively. Transplanting the composition module into five existing memory systems improves accuracy across all five. Beyond these patterns, the search objective shapes where computation goes: optimizing serving cost alongside accuracy shifts computation from LLM-based memory construction to selective processing at query time while retaining competitive accuracy. Together, these findings offer empirical lessons for agent memory design. Code is available at https://github.com/ash-ding/MAPE.","authors":["Xinheng Ding","Seyyed Saeid Cheshmi","Ziyang Zhang","Tairan Chen","Wenya Xie","Yu Hou","Rui Zhang","Zirui Liu"],"cats":["cs.CL","cs.AI"],"primary":"cs.CL","published":"2026-11-12T16:42:04Z","topic":0},{"id":"2611.17067","title":"An Invariant Neural-Network Eddy-Viscosity Closure with Iterative Solver Coupling for Molten Salt Fast Reactor Thermal-Hydraulics","summary":"System thermal-hydraulics codes such as the System Analysis Module (SAM) close turbulence on coarse meshes with algebraic zero-equation mixing-length models. These models cannot represent the non-equilibrium turbulence, flow separation, and large-scale recirculation of open-cavity advanced reactors. We develop an invariant neural-network eddy-viscosity closure for SAM and apply it to the Molten Salt Fast Reactor (MSFR). The closure predicts the turbulent kinetic energy and timescale of each element from the flow and geometry of its neighbors, without solving turbulence transport equations. The resulting non-negative eddy viscosity enters the implicit diffusive operator, which keeps the momentum equations well conditioned. We prove that the closure is exactly invariant to translation, Galilean boost, rotation, reflection, and neighbor ordering. A network trained on high-order Nek5000 fields sees a different feature distribution on the coarse SAM mesh. We remove this covariate shift by iteratively retraining the network on SAM flow states against fixed Nek5000 targets. For a held-out steady state, the coupled closure reproduces the single recirculation cell and eddy-viscosity plume of the Nek5000 reference. In a 120 s pump-coastdown transient, it halves the velocity error of the mixing-length model, from 37.6% to 17.7%, and matches the Nek5000 eddy viscosity with a time-averaged relative error of 12%. The remaining velocity error comes from the coarse mesh. The coupled transient takes about 0.2 core-hours, compared with about 330 core-hours for Nek5000. The results show that the closure can recover two-equation RANS turbulence behavior in a coarse-mesh system code at a small fraction of the CFD cost.","authors":["Zaid Abulawi","William Oliveira","Akzhol Almukhametov","Jun Fang","Ling Zou","Rui Hu","Stefano Lorenzi","Yang Liu"],"cats":["physics.flu-dyn","cs.LG"],"primary":"physics.flu-dyn","published":"2026-11-12T16:41:34Z","topic":6},{"id":"2611.17059","title":"Capacity-Constrained Delayed Feedback for Online Conformal Risk Control","summary":"Online conformal risk control (OCRC) enables reliable uncertainty quantification in machine learning by converting model outputs into prediction sets with long-term risk guarantees. These guarantees, however, hinge on the assumption that the loss of each round is revealed immediately and losslessly. In practice, feedback may be delayed, and limited resources may prevent tracking all pending predictions, causing some feedback to be lost. To address this challenge, we develop capacity-aware OCRC (CA-OCRC), a framework that jointly determines how to allocate limited tracking feedback capacity and how to construct prediction sets. CA-OCRC is based on a randomized tracking policy that samples a proxy delay to decide whether, and for how long, feedback for the current round is tracked. CA-OCRC updates the prediction sets via inverse probability weighting, compensating for the possibility of feedback lost due to delays and limited tracking capacity. We analyze risk bounds for both arbitrary and stochastic delays, revealing a trade-off between receiving more feedback and avoiding capacity saturation. Building on these bounds, we optimize the proxy delay distribution, showing scaling laws for the tracking capacity that ensure the same long-term risk as OCRC with unlimited feedback. Experiments on medical imaging, natural language processing, computer vision, and wireless communications show the effectiveness of CA-OCRC as compared to existing OCRC mechanisms.","authors":["Bowen Wang","Osvaldo Simeone"],"cats":["cs.LG","eess.SP"],"primary":"cs.LG","published":"2026-11-12T16:37:20Z","topic":3},{"id":"2611.17053","title":"Evaluating Pretrained Language Semantics for Structured EHR Prediction Tasks","summary":"Structured Electronic Health Records (EHR) models typically treat each clinical concept as a dataset-specific token, discarding the natural-language semantics associated with the event. As a result, relationships between concepts must largely be learned from EHR co-occurrence, without directly exploiting any prior semantic knowledge. We investigate how pretrained causal language models can be adapted for longitudinal EHR prediction by developing tEHR. Here, we represent events with their natural language description and use this text stream to apply EHR-adaptive continued pretraining, followed by task-specific discriminative fine-tuning using LoRA, utilising the LLM as a longitudinal patient-encoder. We evaluate this approach across four prediction tasks on the MIMIC-IV dataset and pancreatic cancer prediction on the CPRD dataset at 1 to 12 month horizons, with tEHR using the Mistral-7B-v0.3 backbone achieving the strongest overall performance. Ablations show that performance declines when clinical descriptions are removed or misaligned, when numerical values are omitted, and when either stage of adaptation is removed, while explicit elapsed-time descriptions provide little additional benefit. Analysis of the learned patient representations further reveals clinically coherent structure associated with distinct patterns of disease risk and clinical history.","authors":["Ashvin Gupta","Brendan Delaney","Alessandra Russo"],"cats":["cs.CL","cs.LG"],"primary":"cs.CL","published":"2026-11-12T16:34:49Z","topic":4},{"id":"2611.17052","title":"Auditing AI peer reviewers: a benchmark on real scientific papers","summary":"LLM-based peer review is now used at production scale, but its evaluation remains limited. We introduce a benchmark that injects objective, localisable errors of four types (general issues, mathematics, numerics, citations) at three severity levels into real papers and measures each reviewer's detection rate by error type and severity, together with its false-positive rate on clean copies. We also present Skepthical, a multi-agent reviewer that verifies citation-backed claims by retrieving the cited papers, mathematics with computer algebra, and numerical claims with executed code. We apply the benchmark to two corpora of 15 recent arXiv preprints each, one in astrophysics and one in machine learning (ML), injecting a total of 357 errors, and evaluate twelve reviewer systems: six single-LLM reviewers, two Claude Code configurations, three publicly available reviewers, and Skepthical. The choice of model has the largest impact on detection rate: given the same prompt, three frontier LLMs differ by about 50 percentage points in total detection on both corpora. Tools and scaffolding help when they contribute what the model cannot supply on its own. In particular, on astrophysics, Claude Code improves on a single Claude call in mathematics and numerics detection, and enforced citation retrieval gives Skepthical the highest citation error detection rate of all systems by a large margin. This citation lead, however, is not observed on ML, where the models appear to know the literature better and Skepthical's citation detection rate is statistically indistinguishable from that of the best single LLM. False-positive rates are below 10% for most systems, but vary with prompt, model, and scaffolding, with mathematics flags the main source of false positives. The evaluation pipeline, the injected errors, and the adjudication logs are released as a benchmark.","authors":["Íñigo Zubeldia","Boris Bolliet","Francisco Villaescusa-Navarro","Pablo Villanueva-Domingo","Miles Cranmer"],"cats":["cs.CL","astro-ph.IM","cs.AI","cs.DL"],"primary":"cs.CL","published":"2026-11-12T16:33:51Z","topic":0},{"id":"2611.17048","title":"AURA 2.0: From User-Driven Responses to Model-Initiated Conversations in Streaming Video","summary":"Most AI assistants respond only when prompted, remaining silent even when events in the video stream offer meaningful opportunities for conversation. To move beyond purely user-driven interaction, we present AURA 2.0, a streaming video large language model that introduces three core upgrades to AURA. The key advance is Model-Initiated Conversation (MIC), which enables the model to identify topics worth discussing in the evolving visual stream and initiate conversations at appropriate moments without a user query. Stream-Aware Tool Calling (SATC) enables the model to invoke external tools using the current visual and conversational context and incorporate returned information into the ongoing interaction. Interactive Visual Grounding (IVG) supports visual grounding in both user inputs and model outputs. The model can identify objects marked by users on video frames and produce frame-aligned bounding boxes to link its responses to specific objects and regions. To develop these capabilities, we construct a dedicated streaming interaction dataset and adopt a three-stage training scheme to integrate the new behaviors with AURA's existing streaming interaction abilities. We further incorporate Long-Horizon Interaction Memory (LHIM) to maintain contextual continuity over extended video streams. Evaluations on public streaming benchmarks show that AURA 2.0 achieves leading performance among open-source streaming models, while targeted evaluations further validate its new interaction capabilities. We open-source the model weights and inference code.","authors":["Shuhan Li","Huankang Guan","Xintong Guo","Yang Bo","Yuyang He","Dunyuan Xu","Jingyan Chen","Fang Liu","Tsz Kin Wan","Jinpeng Chen","Zhenqi He","Yuxuan Hu","Wenbo Li","Han Gao","Qijun Li","Long Chen","Hongsheng Li","Rui Liu"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T16:31:54Z","topic":3},{"id":"2611.17042","title":"Infrastructure for Verifiable AI: Where Proof Costs Go and How Hardware Can Help","summary":"Users of artificial intelligence services, including those powered by large language models and other transformer-based models, need assurance that inference and training follow the declared computation without exposing private models or data. Zero-knowledge proofs can provide this assurance, but proof generation remains expensive. Hardware acceleration research often reports gains for individual cryptographic operations, leaving their effect on the full cost of proving neural-network workloads, particularly large transformer models, unclear. We introduce a four-layer framework linking verifiable AI workloads, proof systems, cryptographic kernels and hardware platforms, use it to review the literature, and add original measurements of six provers and seven kernels to identify which costs matter in practice. Our measurements show that the dominant cost changes with the proof system and input length. In low-rank fine-tuning, the cost of opening commitments to fixed model weights remains nearly constant across input lengths from 64 to 2048, while the work associated with intermediate values grows. A 13-billion-parameter fine-tuning configuration exceeds the capacity of a 40-gigabyte graphics processor but completes on an 80-gigabyte device, showing why capacity must be assessed alongside execution time. Our kernel profiling further identifies instruction issue and synchronization limits that a bandwidth-only model misses. We use these findings to assess graphics processors, reconfigurable logic, custom chips and existing artificial intelligence accelerators. The resulting research agenda connects proof-friendly numeric formats, reuse of fixed-weight commitments, heterogeneous execution and shared benchmarks to the measured bottlenecks. Effective acceleration requires selecting hardware and proof-system optimizations around the cost profile of the target workload.","authors":["Yihan Xia","Wenxin Xu","Taotao Wang"],"cats":["cs.AR","cs.CR","cs.LG"],"primary":"cs.AR","published":"2026-11-12T16:27:07Z","topic":5},{"id":"2611.17041","title":"From Delegation to Coordination: Learning Efficient Multi-Agent Deep Research","summary":"Multi-agent deep-research systems use delegation to parallelize search and keep long tool traces out of the main agent's context. Yet delegation helps only when task allocation and subtask execution are coordinated: the main agent must issue useful subtasks, while subagents must return reliable evidence. Existing training either optimizes only the main agent or broadcasts the same team advantage to all agents, which can reward misleading subagent reports and penalize useful ones. We introduce LEAD-RL (Learning Effective Agent Delegation with Reinforcement Learning), a joint main-agent-subagent training framework with contribution-aware credit assignment. A training-time rubric evaluates the research utility of each subagent report and gates its inherited team advantage, while final-answer correctness remains the only team reward; no token, step, or speedup reward is used. Across five benchmarks and two model families, LEAD-RL improves answer quality. On BrowseComp, it improves accuracy by 2.49 percentage points and reduces critical-path cost by 24.5% over joint training with shared team credit. During validation, accuracy increases by 6.78 points at approximately constant critical-path cost. These results show that contribution-aware credit assignment improves the quality-efficiency trade-off of multi-agent deep research.","authors":["Yuyang Wu","Yang Liu","Zequn Xie","Junjie Wang","Dan Yang","Zehao Tan","Ding-Chu Zhang","Junming Yang","Yue Shen","Jian Wang","Jinjie Gu"],"cats":["cs.MA","cs.AI"],"primary":"cs.MA","published":"2026-11-12T16:26:27Z","topic":0},{"id":"2611.17035","title":"Prevalence Bounds for Edited Watermarked Text: Power Envelopes and Calibration Sensitivity","summary":"Watermark detectors decide whether one document came from a watermarked language model. Auditors often need a different number: the share of a corpus that did. We study this share for Gumbel-max and green-list (KGW) watermarks when every document may have been edited by an unknown amount. Without restrictions on the edits, the upper end of the identified set is 1. Combining a known least-favourable next-token distribution with a sequential coupling, we derive a power envelope indexed by three budgets: a retention floor, an entropy budget, and a budget for in-place substitutions, which for Gumbel-max make edited positions look less watermarked than human text. The envelope gives an identified set that is sharp for the alternatives it dominates, finite-sample bounds on the realized share of a fixed corpus with either the exact null or a null calibrated on an independent human sample, and confidence sequences for streams. In Gumbel-max simulations where the envelope binds, the bounds cover, and the threshold that maximizes the gap between envelope and null is generally as tight as twelve thresholds. On text from OPT-1.3B and Qwen3-1.7B, 12,000 fresh human documents scored under three keys show no robust departure from the exact null, and all bounds covered in resampling trials from finite pools of documents with random edits, substitutions and paraphrases; at a share of 0.2 the exact-null bound is about [0.14,0.38]. These results are fragile in two ways we make precise. LLM-paraphrased KGW pools fall below the envelope, yet at 2000 documents the bounds still covered in these trials. A simulated null error of 0.01, comparable to the largest deviation we observed at our thresholds, is likewise harmless at 2000 documents but breaks exact-null coverage at 100,000; a null calibrated on an independent human sample keeps coverage there, but needs a large reference sample to stay narrow.","authors":["Kuancheng Ye"],"cats":["stat.ML","cs.CL","cs.CR","cs.LG"],"primary":"stat.ML","published":"2026-11-12T16:21:31Z","topic":1},{"id":"2611.17031","title":"Your Vision Foundation Model Is Secretly an EEG Foundation Model","summary":"Electroencephalography (EEG) foundation models are pretrained on thousands of hours of recordings, yet their frozen features transfer poorly to downstream tasks, and this transfer does not consistently improve with larger models or more pretraining data. We show that a vision foundation model that has never seen EEG is a stronger frozen EEG encoder than these specialized models, once the signals are rendered as images that a vision transformer can read. Our method, SeeEEG, keeps the vision backbone frozen and learns only how EEG is displayed to it. Each channel occupies a dedicated row of patches and is drawn either as a waveform at a fixed gain or as a scalogram whose frequency range spans the patch height, so that task-relevant patterns fall within individual tokens. The montage, filtering, and contrast parameters of this display are initialized from a standard clinical montage and optimized with the task loss through the frozen backbone, which amounts to at most 1,113 trainable parameters per view. We compare SeeEEG with seven EEG foundation models on eight clinical, sleep, affective, and brain-computer interface datasets, with every encoder frozen and paired with identical readout heads. With a frozen DINOv3 ViT-B, SeeEEG outperforms every EEG foundation model in balanced accuracy on all eight datasets and reaches a mean rank of 1.3 in Cohen kappa or AUROC, compared with 2.9 for the strongest baseline, REVE. The learned displays transfer without further optimization to the video encoder V-JEPA 2.1, which outperforms every EEG foundation model on both metrics on six of the eight datasets. Replacing DINOv3 with a randomly initialized ViT-B lowers the score by 15.2 points on average, so the ability to read rendered EEG comes from visual pretraining.","authors":["Guoan Wang","Feng Liu"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T16:17:16Z","topic":3},{"id":"2611.17030","title":"Similar but Not Redundant: Embedding-Gated Refinement for High-Granularity Topic Models","summary":"High-granularity topic modelling extracts hundreds of niche topics from a corpus, yet this regime is rarely studied and is more commonly framed as a failure mode: \"over-clustering,\" in which coherent themes fragment into redundant subtopics. We argue that this framing conflates genuine redundancy with topics that are semantically similar but meaningfully distinct, and that the central challenge in high-granularity modelling is separating the two. Recent post-hoc refinement methods address topic redundancy through either embedding-based similarity thresholding or LLM-guided merging. We show that neither suffices in high-granularity settings and propose embedding-gated iterative merge (EGIM), a hybrid approach that iteratively generates context-rich topic representations, uses embedding similarity to propose candidate merges across the topic space, then applies LLM judgement to accept only genuinely redundant pairs while preserving niche distinctions. Across two patent corpora and one legal corpus, LLM judgement rejects 83% of embedding-identified candidates as non-redundant. EGIM improves discriminability on merged topics by 25-55% over a prior iterative LLM baseline while better preserving the original topic distribution.","authors":["Daniel Denley","Robert L. Nicol","Mike J. Chantler"],"cats":["cs.CL","cs.IR"],"primary":"cs.CL","published":"2026-11-12T16:17:08Z","topic":6},{"id":"2611.17019","title":"Temporal Regularized Learning: Self-Supervised Learning Local in Space and Time","summary":"Layer-local self-supervised objectives remove the backward pass between layers, but almost all still compute their loss over a batch of samples and are therefore local in space but not in time. We introduce Temporal Regularized Learning (TeReL), a spatially and temporally local learning rule for self-supervised representation learning from sequential input streams. The key idea is to instantiate a VICReg-style objective with temporally adjacent positives and non-differentiated running population statistics, yielding a per-neuron online loss with bounded state. The objective is a soft form of Slow Feature Analysis, and its gradient is a Hebbian update of the feedforward weights and an anti-Hebbian update of one lateral matrix per layer. The aim of the experiments is to show that the rule works in a small network, not to compete with backpropagation. On a class-chunked MNIST stream, a two-layer fully connected network trained with batch size one outperforms random features (96.48% versus 95.09% linear-probe accuracy) and classical SFA, and a variant that relaxes both locality constraints is comparable to supervised backpropagation. Because training and inference use the same computation, the network can keep learning from unlabeled data during inference without loss of accuracy.","authors":["Davide Wiest"],"cats":["cs.LG","stat.ML"],"primary":"cs.LG","published":"2026-11-12T16:09:20Z","topic":7},{"id":"2611.17015","title":"AI for Biologists: The Knowledge Boundaries of Foundation Models in the Age of Agentic Systems","summary":"Artificial intelligence (AI) is changing how biological research is done. Biological foundation models trained on millions of nucleotides, protein, and single-cell measurements predict protein structures, mutation effects, and cellular responses to genetic perturbations. Agentic systems, AI assistants built on large language models, can run multistep analyses from plain-language requests and increasingly call foundation models as tools. Yet it is often unclear when these models' outputs can be trusted. We frame this as a model's knowledge boundary: the range of inputs, tasks, and conditions over which its outputs are reliable. This review explains how foundation models are built and proposes that five design elements (data, architecture, training scheme, evaluation scheme, and post-analysis) define this boundary. We examine evidence on where current models succeed or fail and present standardized descriptions of model limits so that researchers, and the agents acting for them, can use these tools reliably.","authors":["Tian Cai","Andrew Kleist","Jaimin Patel","Duccio Malinverni","Alexander K. Taylor","Malvika Sharan","Renato Umeton","M. Madan Babu"],"cats":["q-bio.BM","cs.AI"],"primary":"q-bio.BM","published":"2026-11-12T16:08:33Z","topic":0},{"id":"2611.17011","title":"Random Exploration is Random","summary":"Decision making often requires balancing exploitation of familiar options with exploration of uncertain alternatives. One strategy for exploration is random exploration, in which choices become more variable and less tied to the option that currently appears best. However, it remains unclear whether this variability reflects genuine stochasticity or instead a deterministic structure that traditional cognitive models fail to capture. We examined this question using the Horizon Task in an aggregated dataset of 1,392 participants completing over 1,270,695 trials. In this task, exploration becomes more valuable when participants have more future choices, and choice variability increases under longer horizons. We tested whether more expressive neural network models could recover deterministic structure from this additional variability. Although neural networks provided better overall fits to behavior than the traditional model, long-horizon choices remained consistently less predictable than short-horizon choices. These findings show that limited model expressiveness cannot fully account for the increased variability associated with random exploration, suggesting random exploration truly is substantially random.","authors":["Yingkang Luo","Hua-Dong Xiong","Hanbo Xie","Caroline Phelps","Madeline Harms","Ryan Smith","James Waltz","Robert Wilson"],"cats":["q-bio.NC","cs.LG"],"primary":"q-bio.NC","published":"2026-11-12T16:07:42Z","topic":7},{"id":"2611.17007","title":"Skill Adapter: Personalized Modification and Online Routing of General-Purpose Skills","summary":"Large language model (LLM) agents use external skills to reuse task procedures and experience with tools. General-purpose skills describe shared execution procedures, while specific tool choices, operating practices, and output requirements vary across users. In long-horizon tasks, subgoals, available inputs, and environmental states change as execution progresses. Agents must select currently applicable skills from an expanding skill pool and prepare candidates for subsequent steps. We propose Skill Adapter, which enables personalized modification and online routing of general-purpose skills. The system distills local adjustments from user requirements and execution feedback and stores them as Adapters linked to general-purpose skills. The routing agent selects general-purpose skills and applicable Adapters based on the current subtask, execution progress, and environmental feedback, incrementally maintaining a task-level online skill library that the main LLM reads on demand. The system preloads candidates for subsequent nodes and rechecks their applicability once actual inputs are determined. Reinforcement learning optimizes the two types of decisions using modification gains and routing outcomes, respectively. We evaluate Skill Adapter on ALFWorld, WebShop, and SkillsBench. Compared with SkillRL, it improves task success rates by an average of 11.43 percentage points and reduces mean execution time on all three benchmarks. Skill Adapter retains its performance advantage when the skill pool grows to 1,000 skills.","authors":["Qijia Zhuang","Yanyi Gou","Wenbin Ma","Yuxiang Ren"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T16:05:23Z","topic":0},{"id":"2611.17005","title":"An Integrated Architecture for Data-Efficient Real-Time Neural Imitation of Nonlinear Time-Optimal Model Predictive Control","summary":"Nonlinear time-optimal model predictive control (TOMPC) enables highly performant control of constrained systems, but its computational cost prohibits real-time use on fast dynamical systems. Neural network imitation of the TOMPC control law can reduce online computational costs, yet existing work mostly targets quadratic cost formulations in simulation and addresses the challenges of neural controllers in isolation. This work imitates a long-horizon nonconvex TOMPC with hard terminal constraints, whose high computational cost limits training-data generation. We present (i) \\emph{self-informed sampling (SIFS)}, a data-efficient expert-sampling strategy that uses a learned time-to-go to query the expert only where the policy deviates from time-optimal behavior, (ii) a complete training and control architecture that jointly addresses dataset correlation, suboptimal expert labels, covariate shift, constraint satisfaction and model mismatch, and (iii) real-world validation on a Furuta pendulum. SIFS matches standard dataset aggregation performance in roughly one-third of the training time and with less than half the training data. The resulting controller matches the region of attraction of the expert TOMPC while raising the control frequency from 1.4~Hz to 1~kHz. On hardware, the controller swings up and stabilizes the pendulum along trajectories closely matching the expert TOMPC solution.","authors":["Maarten De Baecke","Flavia Sofia Acerbo","Andrea Giusti","Wilm Decré"],"cats":["eess.SY","cs.LG","cs.RO"],"primary":"eess.SY","published":"2026-11-12T16:04:44Z","topic":4},{"id":"2611.16991","title":"AnalogLayoutWorld: An Open Benchmark for Agentic Analog Layout Design","summary":"Coding agents have demonstrated strong capabilities in software development, yet their ability to tackle long-horizon engineering tasks governed by physical constraints remains less understood. We study this question through analog circuit layout, which demands domain knowledge, geometric reasoning, and iterative interaction with design and verification tools. We introduce \\textbf{AnalogLayoutWorld}, an open-source agent benchmark with 40 expert-reviewed tasks for end-to-end layout generation from fixed transistor-level netlists. It combines an integrated design environment with independent physical and post-layout electrical verification, as well as objective layout quality metrics co-designed with analog layout experts. Across 13 model--harness configurations, 91.1\\% of attempts produce GDS layouts, but only 23.0\\% satisfy the benchmark's acceptance criteria. We also evaluate Novara, an internally developed agent system, which achieves the highest reported aggregate score. Our analysis links performance differences to test-time scaling under fixed budgets, suggesting that reasoning efficiency and effective use of verification feedback help sustain design quality as task complexity grows.","authors":["NovaSilicon Team"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T15:56:13Z","topic":0},{"id":"2611.16986","title":"Primer: An AI Framework for Instructor-Centered Educational Simulation Creation","summary":"Educational simulations are a well-established pedagogical technique that gives learners opportunities to apply knowledge, respond to challenges, and reflect on their performance. High-quality simulations, however, are difficult to build, requiring instructor input, scenario design, character development, scaffolding, technical implementation, and repeated revision. This paper presents Primer, an open-source, LLM-agnostic AI-supported system that helps educators create adaptive educational simulations. Educators contribute their knowledge of the subject, lesson, and students, while generative AI produces responsive variations of practice for students. Primer uses a multi-agent system centered around executing educator's instructional intent through the entire learning experience. We hope that by releasing Primer as an open-source system, including its prompts and GitHub repository, we provide a foundation for development and research on educator-in-the-loop AI authoring, multi-agent educational workflows, and simulation-based pedagogy. Further, by allowing the use of various AI models, including local and open source models, we provide the opportunity for implementation under varied governance regimes. Primer points toward a model of educational AI in which instructors can create locally grounded practice opportunities based on their own materials, goals, examples, and knowledge of students.","authors":["Ethan Mollick","Lilach Mollick","Ben Przystanski","LJ Ciccarelli","Natalie Bach","Laura Zarrow"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T15:50:46Z","topic":0},{"id":"2611.16984","title":"ElasticDino: Dense Semantic Features Using Elastic Deformations","summary":"Vision transformers (ViTs) serve as the backbone for countless vision tasks, including dense prediction tasks, such as segmentation and depth estimation, even though the quadratic complexity of self-attention severely limits feature resolution and induces spatial artifacts. In this work, we introduce ElasticDino, a lightweight adapter which produces high-resolution feature maps, adhering to the fine details of the input image. As a key insight, we observe the consistency of artifacts across channels, and simultaneously resolve them to match the input image structure. Using a slim architecture based on spatial deformations, ElasticDino produces pixel-aligned, high-resolution feature maps efficiently. Leveraging deformations allows strictly preserving sharp semantics, and thereby avoiding the smoothness biases inherent to alternative approaches. Due to the absence of ground truth, we employ pixel-level semantic segmentation and depth estimation with linear heads to predict a set of downstream dense tasks, guiding the deformation process without leaking task-specific semantics into the model. We demonstrate superior accuracy over current methods on a range of popular dense task benchmarks, with better computational performance.","authors":["Ulysse Mizrahi","Sean Zaretzky","Dov Danon","Amit H. Bermano"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T15:50:04Z","topic":3},{"id":"2611.16981","title":"Compiling Known Rules into Compact Neural Agents: Weight-Space Programming on Unseen Games","summary":"Weight-space programming (WSP) uses language models to construct neural network weights without training. We test whether known game mechanics can be compiled into compact agents that generalise to held-out levels, renderings and start states within a game family. Writers receive simulator source with held-out level data removed for 20 privately authored, unpublished ARC-AGI-3-style game families. With a fixed 57k-parameter recurrent architecture, the strongest writer reaches 56% held-out-level success with feedback, while the four main writers' Python players reach 97-100%. An exploratory free-architecture condition reaches 68-90%. Learning from support-level examples transfers much less well, but these learners lack the source information given to writers. Written networks have a program-like structure: 93% of parameters are zero, and 90% of the rest are small integers or powers of two. NumPy players under comparable computation caps reach 91%, versus the networks' 79%. Training the strongest writer's architectures from scratch yields about 1% success within our budget. Written fixed-network weights are fragile, and standard PPO removes most of their success. Prompts for noise robustness increase tolerance by about four orders of magnitude for two of three writers, but PPO makes larger updates and still removes most success. These results demonstrate rule-to-network compilation in this setting, while showing limited benefits over Python programs and no successful combination with the tested PPO fine-tuning.","authors":["Jens Lehmann"],"cats":["cs.AI","cs.LG"],"primary":"cs.AI","published":"2026-11-12T15:49:00Z","topic":0},{"id":"2611.16968","title":"Morphological Simplification in Alzheimer's Disease: Computational Analysis and Identification From Connected Speech","summary":"Purpose: Morphology is often regarded as relatively spared in Alzheimer's disease (AD) and is rarely modeled in connected speech. This study asked which morphological features distinguish speakers with AD from healthy controls, how well morphology classifies AD compared with other linguistic levels and BERT-based models, and whether the signal exceeds transcript length and vocabulary. Method: Fifty-five automatic measures in eight families, seven of them new and theory-motivated, were computed from participant speech in the ADReSS Cookie Theft descriptions (78 AD, 78 controls) and analyzed with group comparisons, equivalence tests, SHAP, family ablation, likelihood-ratio tests, feature-selection classifiers, and BERT-based models. The automatic measures were validated manually, and models were transferred to two other corpora. Results: Nineteen of 38 measures differed between groups: speakers with AD produced fewer suffixed and inflected words, shorter verbs and nouns, and fewer present-progressive but more past-tense forms (largest effect r = -0.42), whereas errors were rare and statistically equivalent between groups. Eight measures survived adjustment for length and lexical frequency. With test-set feature selection, morphology reached a macro-F1 of 0.85, on par with syntax and above fine-tuned BERT (0.80 over five seeds); fully cross-validated selection gave 0.62. Adding morphology to BERT improved AUC (0.91 vs. 0.89) but not macro-F1. Transfer was limited, although two thirds of the effects kept their direction in mild cognitive impairment. Conclusions: AD connected speech shows morphological simplification and temporal reanchoring rather than breakdown. Morphology is more informative than its reputation suggests. Keywords: Alzheimer's disease; morphology; connected speech; ADReSS; inflection; tense and aspect; SHAP; machine learning; BERT","authors":["Mohamed Zakaria Kurdi"],"cats":["cs.CL"],"primary":"cs.CL","published":"2026-11-12T15:36:47Z","topic":3},{"id":"2611.16961","title":"The Value of Thinking: Learning When to Reason in Hybrid Decision Models","summary":"Existing adaptive-reasoning methods shorten the explicit reasoning a model writes, or skip it by switching modes; either way the answer is still written out token by token. \\emph{Rather than deciding how much to think}, we explore a different paradigm, \\emph{zero-or-$T$ thinking}: a single model either answers from the implicit thinking inside one forward pass, generating no token, or performs full autoregressive reasoning at a cost of $T$ tokens when that is worth paying. We realise it in a \\emph{hybrid decision model} (HDM), where one frozen backbone serves both modes and a lightweight \\emph{self-routing block} chooses between them. The block estimates the \\emph{value of thinking}, the utility the slow mode is expected to add over the fast one, from what that implicit thinking has already produced, and the slow mode runs only where that value exceeds a price $λ$ set per request. Where no per-decision label exists, as in episodic agentic tasks, we introduce \\emph{Calibrated Decision Policy Optimisation} (CDPO), a reinforcement-learning recipe that trains the block from episode-level returns. We evaluate HDM on MMLU-Pro, AGIEval, GPQA-Diamond, JevBench, BANKING77 and MiniWoB. Specifically, on JevBench HDM-4B scores $0.857$ against Jev's $0.851$ ($+0.7\\%$) at a median $0.179$\\,s against $0.652$ ($-72.5\\%$). On the agentic venue MiniWoB HDM-27B cuts token cost by $80.1\\%$.","authors":["Zhengxu Yu","Ke Xu","Meng Fang","Haitham Ammar","Weilin Luo","Jun Wang"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T15:32:47Z","topic":0},{"id":"2611.16956","title":"Lara Mini Base: Language Diversity as a Resource, Not a Trade-off","summary":"General-purpose language models are trained mostly on English: over open models we review, at most, non-English pretraining data makes up only a 12% share. Models built for European languages give them far more, but score well below general-purpose models of the same size. A common concern is that, in small models, languages compete for capacity. We present Lara Mini base, a model trained from scratch on 6T tokens in English and 44 other languages, which together cover all 24 official languages of the European Union. As training progresses, the 44 other languages together come to outweigh English. To give these languages enough capacity at low cost, Lara Mini base uses a mixture of experts architecture with 7.2B total and 0.6B active parameters. Among open base models with up to 2B active parameters, it achieves the highest average score across standard English, multilingual, math, and code benchmarks, ahead of models trained on 4 to 6 times as many tokens. Its margin is largest on multilingual benchmarks in the languages it supports and on math benchmarks. It supports contexts of up to 128k tokens. We find that our MoE architecture reduces capacity competition between languages. For a single user on one GPU, without quantization, it generates text at least 1.8 times faster than any model with a higher average score. We release the weights of Lara Mini base under the CC BY-NC-SA 4.0 license. In addition, we will release T-HellaSwag, a higher-quality machine translation of the commonsense reasoning benchmark HellaSwag into 110 language variants.","authors":["Translated","Maria Pia Adinolfi","Isabelle Andrieu","Grazia Fiorella Anfuso","Melania Annunziata","Daniele Antonini","Marta Arena","Luca Barone","Dana Barras Franco","Francesca Basiricò","Francesca Bee","Sofia Bertini","Nicola Bertoldi","Giorgio Biancini","Alessandra Bigi","Giulia Boccalini","Tommaso Bonomo","Enrico Boscardi","Sebastien Bratieres","Francesco Brigante","Giulia Brignola","Romolo Buondestino","Brenda Patricia Byrne","Giorgia Campora","Edoardo Cappelli","Veronica Carascosi","Francesca Carducci","Cody Carlson","Davide Caroselli","Ambra Cascio","Dario Cascio","Mauro Cassani","Andrea Castro","Alessandro Cattelan","Giulia Ceccacci","Paola Cellini","Giulia Chiriatti","Federico Cialini","Valentina Ciarlariello","Daniele Cocozza","Emanuela Coghe","Kathleen Compares","Marco Comunità","Davide Conti","Helen Rebecca Corfield","Michele Cortellessa","Anna Cracco","Aurora Cuppone","Arianna D'Andrea","Maurizio Davoli","Andrea De Angelis","Andrea De Blasi","Rebecca De Giorgi","Annalisa De Petra","Arianna De Pilla","Joris Demiraj","Elena Deufemia","Federica Di Antonio","Pierluigi Di Cianni","Vincenzo Di Cicco","Michela Di Clemente","Alan Di Francesco","Ylenia Di Gese","Federica Di Iulio","Claudia Di Lorenzo","Andrea Di Paola","Giulia Di Pilla","Jessica Dickson","Francesco Elefante","Ilaria Fabrizio","Alessandro Falappa","Viviana Fantozzi","Antonio Farina","Andrea Fernandez Tostado Ro. Diego Andrea","Andrea Ferrari","Sara Ferri","Simona Ferri","Fabio Fioramonti","Stefano Florio","Mariachiara Fornaro","Anna Maria Foti","Niccolò Fransoni","Alessandro Fusacchia","Emma Gamba","Francesco Gasperoni","Davide Gattini","Marta Giani","Claudia Giannetti","Valerio Giannini","Antonella Giudici","Mara Gramendola","Michele Grandinetti","Eleonora Graziani","L'Meese Hope Greaney","Dario Gubernari","Filomena Guglielmi","Jonathan Hammond","Stephen William Harris","Rowena Elizabeth Hogg","Marta Hontanon Del Olmo","Marco Ilari","Kenneth Larry Jackson","Sagar Joshi","Bartosz Alberto Karman","Iuliia Khimenko","Valerio Lanni","Giulio Leonardi","Veronica Levco","Matteo Licci","Alessio Ligios","Noemi Lomartire","Domenico Lupinetti","Jennifer Mirjam Maechtlen","Chiara Maggiori","Maria Luisa Malerba","Francesco Marconi","Domenico Marini","Francesca Marsili","Andrea Martinelli","James Mc Groary","Matteo Menicacci","Andrew Steven Miller","Marco Moauro","Giuliana Molon","Ilaria Mongelluzzi","Emiliano Morghen","Giovanni Moscati","Filippo Musmeci","Flavia Napoleoni","Nicola Giuseppe Ninni","Beatrice Ottoni","Federica Pagliani","Nicolò Palmiero","Daniele Palumbo","Chiara Pantani","Stefano Paolucci","Deborah Paritario","Michela Pascale","Thomas Passera","Daniele Patrioli","Chiara Patrizi","Silvia Patrizi","Giulia Pelos","Simone Giovanni Perone","Cristiano Perrone","Joshua Peterson","Katia Piredda","Francesca Piscitelli","Alessandra Posca","Lorenzo Pratesi","Francesca Prencipe","Daniela Proietto Cuco","Mattia Pupino Marangio","Disheng Qiu","Mirco Ramo","Federica Rapagnà","Waldo Luis Ribeiro","Federico Ricciuti","Alessia Ridoni","Eleonora Risi Valdettaro","Jon Matthew Ritzdorf","Serena Rivera","Federico Roma","Arianna Ruggiero","Eleonora Ruggiu","Mariafrancesca Sagliocco","Vanessa Salvi","Alessio Sampieri","Elena Santanelli","Roberto Sastre","Luca Scofano","Alessandro Sdei","Giuseppe Sebastiani","Claudia Nicole Siegenthaler","Monica Signoretti","Nicola Sobieski","Biancamaria Soellner","Valentina Stati","Michael Stevens","Arianna Stissi","Fabiana Stomaci","Leonardo Taddei","Christopher Thomas","Maurizio Tiberi","John Tinsley","Riccardo Tittarelli","Carla Tocco","Martina Totaro","Barbara Veronika Totzauer","Alessio Tranzocchi","Letizia Treossi","Mara Trombetti","Marco Trombetti","Andrea Valenzi","Kirti Vashee","Giulia Vestita","Teodosio Vezza","Giulia Vignola","Sara Virtuoso","Valentina Volpini","Francesca Vurro","Jolita Zaksauskaite","Giulia Zanarella","Paolo Zanotti","Francesca Zavettieri","Irene Zorzi"],"cats":["cs.CL","cs.AI"],"primary":"cs.CL","published":"2026-11-12T15:28:56Z","topic":4},{"id":"2611.16937","title":"S1-Evo: Experience-Guided Evidence Competition Attacks on Typed System-1 Decision Models","summary":"Typed System-1 (S1) decision models are increasingly deployed in LLM agents for routing, moderation, and authorization to cut inference cost, yet untrusted context can still manipulate typed outputs and alter execution branches. Unlike autoregressive System-2 (S2) models, S1 interfaces score dynamic rules and options without token-by-token decoding, mismatching attacks targeting text continuation or explicit override commands. We propose S1-Evo, a self-evolving framework tailored to S1's contract-based option competition and margin feedback. S1-Evo perturbs untrusted fields via contract-aware evidence channels to induce \\emph{evidence competition} with original task evidence while preserving fact invariance. Offline on seed tasks, it optimizes candidates via feedback search guided by directional margins and re-verification, distilling verified flips into experience cards. At test time, S1-Evo retrieves relevant cards to construct perturbations for unseen tasks. Across four benchmarks and three S1 models, S1-Evo reaches 71.99\\% mean ASR and 57.47\\% high-confidence ASR on \\texttt{jev-1.14}, and causes up to 24.00\\% E2E agent violations.","authors":["Kaixiang Wang","Junjie Li","Zhaojiacheng Zhou","Jiong Lou","Jie Li"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T15:21:44Z","topic":0},{"id":"2611.16935","title":"Universal Infectious Disease Forecasting with Mantis 2","summary":"Public health agencies need to anticipate how outbreaks will unfold, when they will peak, and how long demand on health services will remain elevated. We previously developed a foundation model, Mantis, which was trained entirely on simulated outbreaks---enabling us to generate the diversity and scale of training data needed to support forecasting across diseases. However, our previous work left unresolved when simulation improves predictions and how it should complement observational data. Here, we varied the balance of simulated and observed training data while holding forecasting tasks fixed and varying the available surveillance history. We found that simulation-rich training was particularly valuable with limited local history, and its relative advantage diminished as more observations became available. We also found that simulation-based training reduced exceptionally large forecast errors, suggesting that mechanistic knowledge can help compensate for limited local evidence. Building on these findings, we developed Mantis 2, which combines simulated and observational training to jointly forecast up to six variables sixteen weeks ahead. Across 30 forecasting tasks, Mantis 2 achieved the best aggregate probabilistic accuracy over short, medium, and long horizons among the machine learning, statistical, and baseline models evaluated. It missed roughly 30\\% fewer outbreak peaks than the strongest comparison model, with successfully anticipated peaks identified more than six weeks in advance on average. We publicly release Mantis 2, enabling public health agencies and researchers to forecast disease outcomes without developing a separate model for each disease or location.","authors":["Carson Dudley","Reiden Magdaleno","Yutong Bi","Anthony Valencia","Xiaofeng Liu","Marisa Eisenberg"],"cats":["cs.AI","q-bio.PE"],"primary":"cs.AI","published":"2026-11-12T15:20:49Z","topic":7},{"id":"2611.16934","title":"Time Flies When You Parallelise: Pareto-efficient Time Series Foundation Models with Linear Recurrences","summary":"Time series foundation models have demonstrated strong zero-shot forecasting performance, but computational efficiency remains a central architectural challenge. Temporal self-attention incurs quadratic cost in the number of temporal tokens, while nonlinear recurrent state transitions restrict parallel computation along the temporal dimension. We introduce ci-proto, a family of time series foundation models based on Mamba-3, whose linear state recurrences support chunk-wise parallel-in-time computation during training, and inference costs that scale linearly with sequence length. The architecture alternates recurrent temporal processing with cross-variate attention, supporting multivariate probabilistic forecasting with past-only and future-known covariates. Across fev-bench's 100 forecasting tasks, ci-proto XL sets a new state of the art among time series foundation models. The five ci-proto variants also make up the entire Pareto frontier for mean scaled quantile loss and mean absolute scaled error against both parameter count and median runtime.","authors":["Torben Berndt","Lingyi Yang","Federico Nardi","Yi Heng Lim","AJ Shears","Benjamin Walker"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T15:20:31Z","topic":4},{"id":"2611.16933","title":"Kezek: Splitting Mono Two-Person Conversations into Per-Speaker Tracks for Full-Duplex Dialogue Data","summary":"Full-duplex spoken dialogue models learn turn-taking, overlaps and backchannels from conversations recorded with one channel per speaker, but most available conversational audio is a single mono mix. We present Kezek, a model that turns a mono recording of exactly two people into one track per person with the original length and timing; overlaps and short replies (\"yeah\", \"mm-hm\") stay on the track of the person who said them. Kezek decides who speaks when and who is who once for the whole recording (stage 1: a two-voice activity model guided by a speaker fingerprint trained to be insensitive to phase), and a conditioned time-domain separator only carries out that decision (stage 2), with outputs that always sum to the input. Standard separation metrics miss the failures that matter for dialogue data, so we report short-reply retention, leakage onto the other person's solo, channel swaps and coverage. Against the open pipeline pyannote -> DialogueSidon on the same recordings, Kezek keeps more short replies (0.83 vs. 0.69 on Open Yap test; 0.80 vs. 0.71 over the other's speech on unseen telephone speech from CallHome English), mostly because it returns the whole recording while the external pipeline drops part of it (on the audio both return, reply retention is about the same: 0.859 vs. 0.865 on Open Yap); it also leaks less, swaps channels less often, and runs 4-5 times faster (about 230 hours of audio per GPU-hour). We also report negative results and an ideal-mask ceiling that locate the remaining gap: quiet replies under a louder voice. Code and weights are released under permissive licences.","authors":["Bagzhan Karl","Alizhan Karl"],"cats":["eess.AS","cs.AI"],"primary":"eess.AS","published":"2026-11-12T15:19:35Z","topic":3},{"id":"2611.16929","title":"SPADE: Soft Prompts for Ai-text DEtection","summary":"The easy access to Large Language Models (LLMs) has made AI-generated texts on the web increasingly ubiquitous, making AI-text detection a crucial task. Prompts greatly influence the generation process, but existing detectors focus solely on the generated texts --- supervised methods learn from labeled texts only, and zero-shot methods rely on the text's probability distributions --- ignoring the underlying generation process, which limits their robustness in open-world scenarios. We introduce SPADE: Soft Prompts for Ai-text DEtection, a novel framework for detecting AI-generated text that transfers better across datasets. Specifically, we measure the prompt's influence and use it as a discriminative signal. We characterize the prompt's influence by introducing a Pointwise Mutual Information (PMI) term that captures how much a prompt explains a text. Based on this quantity, SPADE trains one soft prompt per class to amplify the separation, and predicts the class whose prompt best explains the candidate text. Extensive experiments on several benchmarks against various detectors demonstrate the robustness and effectiveness of our method, achieving state-of-the-art (SOTA) performance on out-of-distribution (OOD) datasets.","authors":["Alex Jiang","Didier Schwab","Chuyuan Li"],"cats":["cs.CL","cs.AI"],"primary":"cs.CL","published":"2026-11-12T15:16:19Z","topic":6},{"id":"2611.16928","title":"STEALTH: Steganographic Text Encoding Agnostic to Low-Level Tensor Hardware","summary":"Provably secure linguistic steganography hides a message inside text sampled from a language model, so that without the key, a stegotext is computationally indistinguishable from an honest model sample. Nearly every such scheme assumes the receiver can recompute the sender's token probabilities exactly bit for bit. We show this assumption is fragile: across three model families, bit-exact schemes decode almost none of our messages once the receiver's floating-point precision, quantization, or GPU architecture differs from the sender's, or even when the receiver merely scores a text in parallel versus one token at a time, with the same model, precision, and hardware held fixed. We present STEALTH, a scheme built for exactly this setting. The sender draws each token with an exponential race driven by keyed pseudorandom noise, so a receiver's probability mismatch flips a token only when the race is close; a rateless spinal code, decoded by beam search, absorbs the resulting flips without the sender and receiver ever agreeing on a single distribution. We prove stegotexts remain indistinguishable from honest samples under this construction. Empirically, STEALTH decodes every message we sent for every receiver in the sender's model family, at every precision and quantization level, on smaller models, and on a different GPU architecture. STEALTH is even robust across different model families with separate vocabularies and tokenizers, decoding every message we tested. Our approach carries far more payload per token than prior robust schemes, and trained steganalyzers and a targeted statistical test both fail to distinguish its stegotexts from honest model output.","authors":["William Fleshman"],"cats":["cs.CR","cs.AI","cs.LG"],"primary":"cs.CR","published":"2026-11-12T15:15:43Z","topic":6},{"id":"2611.16919","title":"Beyond Word Error Rate: A Failure Taxonomy of Streaming Dictation ASR","summary":"Speech recognition (ASR) systems are usually compared by word error rate (WER) on recorded files, after punctuation, capital letters and number format are removed. In dictation, the audio is streamed live and the text goes into the user's message as written, so what matters is which mistakes the user has to fix. The test set, streamed to eleven ASR systems, comprises 106 scripted English dictations from one speaker (read in a quiet room, with a fan on, and whispered), 20 clips in each of eleven other language and dialect sets, and 400 recordings that join clips in two languages. Every error was sorted by a fixed rule into four levels: recognition, output formatting, language handling and, for the self-hosted models, audio dropped before the model. Systems with similar WER failed in different ways. Zoom Scribe and OpenAI's GPT-Live-Transcribe have almost the same WER on quiet English (7.4% and 7.3%), but 34.7% and 8.8% on whispered English. Zoom and both AssemblyAI models left out 26-28% of whispered words, and still 21-26% with their end-of-turn detection relaxed or off; seven other systems left out at most 3%. Qwen3-ASR's 32% came from the speech detection in our own deployment (6% when relaxed). Across 107 EARS speakers, the same four systems lost 7-31% of whispered words, while six others lost under 2%. With only English selected, Gemini, Deepgram and our self-hosted Whisper lost the non-English part of a two-language recording in 86-99% of recordings, against 33-46% with both languages selected, while GPT-Live-Transcribe, OpenAI's realtime Whisper and Meta lost it in at most 20%. Running Whisper behind a Silero voice activity detector that drops audio judged to be silence raised whispered WER from 10.1% to 51.4%. Most English results come from one read speaker. The code, the recordings, both references and every transcript are released.","authors":["Harras Mansoor","Muhammad Wisal","Hassan Bhatti","Masood Shakir"],"cats":["cs.CL","cs.SD"],"primary":"cs.CL","published":"2026-11-12T15:09:15Z","topic":3},{"id":"2611.16913","title":"CADET: Construction of PCB Routing Instances with Executable Demonstrations","summary":"Printed circuit board (PCB) routing is the stage of physical layout that finds paths connecting the placed pads of each net. Public datasets in this domain are few, and most contain only the final states of routed boards, not demonstrations of the electronic design automation (EDA) engine actions that produce them. To close this gap, we propose CADET (Constructing executAble Demonstrations for Engine-grounded Training), which jointly generates synthetic boards and demonstrations built from EDA engine actions. The resulting RouteDemos provides 100,000 synthetic boards routed with zero design rule violations (DRVs), together with their demonstrations. Their distributions over 17 statistics resemble those of 142 real reference boards. CADET trains a routing policy on these demonstrations with behavior cloning and fine-tunes it with reinforcement learning (RL). On real test boards, the policy raised the rate of complete DRV-free routing from 38.0% to 67.2% over an RL-only baseline. This rate was comparable to that of the open-source autorouter FreeRouting and exceeded those of the other baselines, including GPT-6 Astra. We release the dataset and its generation code as open source at https://github.com/LGAI-Research/CADET.","authors":["Hyungseok Song","Won-Seok Choi","Seohui Bae","Han-Seul Jeong","Junseok Park","Youngjoon Park","Soonyoung Lee"],"cats":["cs.AI","cs.LG"],"primary":"cs.AI","published":"2026-11-12T15:03:50Z","topic":2},{"id":"2611.16905","title":"Factored One-sided Adam-Moment for Practical and Scalable SOAP","summary":"Second-order optimizers (e.g., Shampoo, SOAP) have emerged as compelling alternatives to Adam/AdamW and promise faster per-step convergence than AdamW in transformer training. However, they still remain costly in practice, and even SOAP still pays substantial optimizer-state memory and eigendecomposition overhead. We ask whether the benefits of SOAP can be retained at nearly the efficiency of AdamW. In this paper, we answer yes with a lightweight SOAP variant, Factored One-sided Adam-Moment (FOAM). The path to practical impact is twofold: first, FOAM employs input-only preconditioning, dropping the output-side Kronecker factor to reduce memory and remove one eigendecomposition. The Kronecker structure of the Hessian shows that this is exact for a diagonal output-side curvature and nearly lossless for a near-diagonal one, as we observe empirically. Second, FOAM replaces SOAP's dense rotated second moment with a factored Adam-moment estimator based on row-by-column statistics. Under the K-FAC Fisher assumption with a diagonal output curvature, this estimator exactly recovers SOAP's dense second moment, and we measure only a 15-20% K-FAC violation, with negligible effect on the update. Our empirical results support the claim that FOAM matches strong SOAP and Muon baselines in validation loss across scales of large language model (LLM) pre-training, uses half of SOAP's optimizer state, runs at Muon-class wall-clock time, and retains robustness across a three-fold learning-rate range. The results hold from 124M to 760M parameters, demonstrating that SOAP's gains need not require SOAP's cost. Code: https://github.com/naver-ai/foam","authors":["Jaemyung Yu","Byeongho Heo","Sangdoo Yun","Dongyoon Han"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T14:59:01Z","topic":4},{"id":"2611.16903","title":"Uncertainty-aware quantum machine learning based on Dempster-Shafer theory","summary":"A large body of work in quantum machine learning has centered on characterizing the expressivity and performance of quantum models. In contrast, little attention has been paid to the reliability of predictions beyond average performance measures such as accuracy. In this work, we introduce an uncertainty-aware quantum classification framework based on Dempster-Shafer theory. This theory allows to express uncertainty through weights assigned not only to individual classification labels but to sets of labels. We introduce a training strategy that enables parameterized quantum circuits to learn such assignments directly from data. This provides an efficient way to prepare the corresponding quantum states and enables the application of different post-processing strategies. In particular, it offers a practical setting for quantum Dempster-Shafer fusion procedures with polynomial resource requirements, even when the corresponding mass functions are defined over exponentially many subsets. Overall, our work represents a step toward enabling quantum models to express uncertainty about their predictions, thereby extending the evaluation of quantum models beyond predictive performance alone.","authors":["Marie Kempkes","Jakob Spiegelberg","Evert van Nieuwenburg","Vedran Dunjko"],"cats":["quant-ph","cs.LG","stat.ML"],"primary":"quant-ph","published":"2026-11-12T14:58:31Z","topic":4},{"id":"2611.16895","title":"Human or Synthetic? What a Fixed Budget Buys for Post-Training","summary":"Post-training of large language models typically combines supervised fine-tuning (SFT) with preference-based fine-tuning such as Direct Preference Optimization (DPO). Human-annotated and synthetically generated preference data differ substantially in per-sample cost, but prior work typically holds sample count fixed when comparing the two. In this work, we explicitly study the interplay between cost and performance by comparing the effect of post-training with human and synthetic preference data under a fixed, dollar-matched annotation budget. We train and evaluate multiple base model families across helpfulness and summarization tasks. We find that synthetic data is far more cost-efficient at low-to-moderate budgets, achieving high win rates with as little as \\$200--\\$400. On helpfulness, human data closes the gap only at budgets roughly two orders of magnitude larger; on summarization, it does not close the gap at any tested budget. We also find that human-annotated data varies in reliability: one widely used human preference source substantially underperforms another, and the gap persists even after standard remedies such as domain restriction and deduplication.","authors":["Shourik Banerjee","Ethan Mendes","Mohit Raghavendra","Wei Xu","Alan Ritter"],"cats":["cs.CL","cs.LG"],"primary":"cs.CL","published":"2026-11-12T14:51:51Z","topic":4},{"id":"2611.16890","title":"Look It Up: Multi-bit LLM Watermarking via Codebook Search","summary":"Attributing short LLM-generated texts to a specific source remains challenging for existing multi-bit watermarking methods, which often require long generations or costly decoding to recover payloads reliably. We propose a codebook-based watermarking scheme that maps an $n$-bit payload to a longer $u$-bit codeword and embeds it during generation through token-level logit biasing. During extraction, the received bit sequence is matched to the nearest codeword in a precomputed secret codebook using Hamming or Levenshtein distance. Experiments show that our method achieves near-perfect recovery for 16-24 bit payloads on 200-token generations and strong performance on 127-token generations for base models. Results are weaker on some instruction-tuned models with very short texts, where low-entropy output distributions reduce embedding reliability. We also study a BCH-based variant that avoids codebook storage but generally reduces robustness. The Levenshtein variant improves resilience to deletions and light paraphrasing, while insertions and stronger rewrites remain challenging. We analyze collision probability, codeword separation, extraction time, tokenization consistency, and text quality, highlighting both the practicality and limitations of codebook-based watermarking for short-form attribution.","authors":["Jarosław Janas","Paweł Morawiecki","Josef Pieprzyk"],"cats":["cs.CR","cs.AI","cs.CL"],"primary":"cs.CR","published":"2026-11-12T14:49:23Z","topic":1},{"id":"2611.16886","title":"LLM-Assisted Verification in Dafny","summary":"Formal verification of complex algorithms requires expertise in program specification, invariant construction, mathematical proof, and the management of automated theorem provers. Although large language models (LLMs) can generate program code and proof fragments, reliably developing complete machine-checked verifications remains challenging: it requires an iterative process of constructing and validating specifications and invariants, building proofs, diagnosing solver failures, and optimizing verification performance. This paper presents a systematic ten-step workflow for LLM-assisted verification of scientific algorithms using the Dafny verification system. The workflow integrates algorithm formalization, contract development, empirical invariant validation, proof construction, bottleneck profiling, and resource-driven proof optimization. We applied the workflow using Antrhopic's Claude Opus 5 and Opus 5.5 models, combining LLM-driven development with automated testing and verification feedback, while retaining human oversight for correctness-critical decisions and strategic interventions. We evaluate the approach on 37 complex algorithms from graph theory and labelled transition systems. Notably, we verified the Hopfcroft-Tarjan triconnectivity test, demonstrating that even highly challenging algorithmic verification tasks can be carried out through a structured LLM-assisted workflow. Many of our machine-checked verifications were obtained within hours to a few days of interactive development. Our results suggest that LLM-assisted workflows can shift the role of the human verifier from detailed proof construction toward supervision, specification review, and strategic guidance.","authors":["Wieger Wesselink","Kees Huizing","Huub van de Wetering"],"cats":["cs.LO","cs.AI","cs.SE"],"primary":"cs.LO","published":"2026-11-12T14:48:38Z","topic":6},{"id":"2611.16882","title":"Pretraining Data Statistics Shape the Phases of Learning Entity Comparison in Language Models","summary":"How does data shape language model (LM) behavior throughout pretraining? We investigate this question through a case study on entity comparison, e.g., Between France and Brazil, which country is larger?, across nine comparison tasks spanning five entity types. We begin with controlled experiments in which we train small LMs (163M parameters) on mixtures of natural text from pretraining corpora and synthetic task data. Our analysis identifies three distinct phases of learning: (1) an early phase where the LM selects entities by frequency, (2) a middle phase where the LM selects entities by position in the prompt (first vs. last mentioned entity), and (3) a late phase where the LM selects the entity that is the correct answer. We show that the emergence of these three phases is causally determined by statistical properties of the training data. With small amounts of task-specific synthetic data, we observe only the first two phases and the model fails to learn the task; with large amounts, the model jumps directly from the frequency-based heuristic to solving the task correctly. Moreover, changing the entity frequency distribution from a naturally occurring Zipfian distribution (a small number of entities are very common and the vast majority are rare) to a uniform one weakens the frequency phase and speeds up learning of the task. Lastly, we show that the three phases of training dynamics from our controlled experiments appear in the open source model OLMo-2 32B. Together, our findings demonstrate that properties of pretraining data are causal drivers of heuristic learning and show that small-scale synthetic experiments can predict training dynamics at larger scales.","authors":["Yik Siu Chan","Jing Huang","Yanai Elazar","Atticus Geiger"],"cats":["cs.CL","cs.LG"],"primary":"cs.CL","published":"2026-11-12T14:45:34Z","topic":6},{"id":"2611.16881","title":"Protein Language Models Fail To Capture Contact Predictions Under Invariant Relabeling","summary":"Protein language models (pLMs) produce highly accurate contact maps from single-sequence inputs. People often take this accuracy as proof that these models implicitly learn invariant coevolutionary statistics, extracting the same signals as classical alignment-based methods like Potts. In this work, we introduce a diagnostic framework around a global bijective relabeling of the amino-acid alphabet. This transformation permutes residue letters while leaving all alignment-level information, column entropies, and pairwise edit distances unchanged. Pseudolikelihood Potts models and inverse-covariance estimators remain invariant under this transformation, but pLMs fail the test. Across multiple architectures (ESM-2, MSA Transformer, AMPLIFY) and model scales (8M to 3B parameters), pLM contact maps collapse under arbitrary relabeling. Using Hamming distance controls and biochemical grouping interventions, we attempt to isolate what likely causes this degradation and find that these models rely on explicit, canonical residue identities (``alphabet anchoring'') rather than pure coevolutionary statistics. Increasing scale makes models more responsive to biochemical classes, but their performance still degrades under relabeling. Our findings show a gap between how pLMs and classical coevolutionary models compute contacts. More broadly, our work shows how diagnostic perturbations that leave classical models provably unchanged can expose hidden representation biases and prevent over-reliance on neural black boxes in biological modeling.","authors":["Paarth Batra","Aryan Keluskar"],"cats":["q-bio.QM","cs.AI"],"primary":"q-bio.QM","published":"2026-11-12T14:45:07Z","topic":6},{"id":"2611.16879","title":"Context Graph: Explainable Multi-Agent Decision Chains in AI-RAN","summary":"AI-RAN distributes network control across multiple autonomous agents, namely xApps and rApps operating over shared radio resources. When an alert or degradation appears, operators currently lack a structured way to trace which decision caused it, what context triggered that decision, and how other agents interacted. Existing explainability tools focus on single-model feature attribution or linear event logging, neither of which captures cross-agent decision chains. We propose the Context Graph, a heterogeneous directed graph that organizes policies, measurements, decisions, and alerts into a queryable explanation layer. Its foundational primitive is the Decision-Event node acting as a causal junction, where incoming edges capture triggering context and outgoing edges track downstream consequences that are progressively upgraded through counterfactual verification. Instantiated over existing E2, A1, and O1 data flows, our proof-of-concept demonstrates that compliance and diagnosis queries are resolved at sub-millisecond latency over graphs of tens of thousands of nodes, while reliably distinguishing true causal impacts from spurious correlations.","authors":["Xijun Wang","Zhaoyang Liu","Xiang Chen","Howard H. Yang","Chenyuan Feng"],"cats":["cs.NI","cs.AI","cs.IT"],"primary":"cs.NI","published":"2026-11-12T14:43:22Z","topic":0},{"id":"2611.16874","title":"Stochastic Kernel Mode Decomposition: Information, Uncertainty, and Conditional Regime Inference","summary":"When do mode energies preserve regime information as measurement quality changes? Stochastic Kernel Mode Decomposition (sKMD) fits joint regime laws on signed KMD mode coordinates and transfers the reference mode posterior to regime probabilities and conditional reconstructions. For centered Gaussian regimes with independently varying known noise variances, fixed coordinate energies retain the full regime posterior precisely when all classes share their cross-covariances and class variances can differ only at isolated vertices of the covariance graph. We extend the characterization to nonzero means and show that freely varying correlated noise forces a fixed linear log-odds decoder to retain the full quadratic feature span even for two distinct centered regimes. At fixed quality, a three-regime example has no sufficient two-energy basis. Its smallest Bayes log-loss gap over bases of condition number at most $κ$ has sharp order $κ^{-1}$. Exact Bayes energy posteriors supply numerical gap estimates for specified bases. Additional bounds relate feature compression, independently learned mode spans and sensor-noise error to log loss. Two controlled experiments with 64 independent panels each show that signed information can improve prediction over energy summaries, while integration gives small gains over strong signed controls and depends on how the competing point rule is calibrated and how accurately observation noise is measured.","authors":["Boumediene Hamzi","Houman Owhadi"],"cats":["stat.ML","cs.CE","cs.LG","math.DS","nlin.PS"],"primary":"stat.ML","published":"2026-11-12T14:39:48Z","topic":4},{"id":"2611.16871","title":"DPA: Decentralized Primal Averaging with Quasi-Global Momentum for Highly Heterogeneous Data","summary":"Decentralized training of deep learning models keeps training data local and supports on-device learning over networks. In realistic scenarios, heterogeneity across agents' local data poses an optimization challenge and can severely degrade test accuracy, especially on sparse communication graphs. Building on recent advances in centralized primal averaging, we propose Decentralized Primal Averaging (DPA), a decentralized learning algorithm that combines primal averaging with a local quasi-global momentum buffer. DPA maintains three coupled sequences for optimization, gradient evaluation, and output averaging, while using consecutive mixed iterates to construct a network-aware momentum direction without transmitting an auxiliary tracking variable. For smooth nonconvex objectives, we establish a convergence bound for DPA and identify conditions for linear speedup in the number of agents for sufficiently many iterations. We also introduce DPA-1G, a variant that uses one communication round per iteration and matches the per-iteration communication cost of DSGD-style baselines. Experiments following the GUT decentralized image-classification benchmark show that DPA compares favorably with DSGD, gradient tracking, QG-DSGDm, momentum tracking, and QG-GUTm across datasets, architectures, topologies, and heterogeneity levels, with the largest gains under high heterogeneity on sparse rings. At the same total communication cost, both DPA variants achieve higher mean test accuracy than the best baseline on CIFAR-10 and CIFAR-100 with ResNet-20. Our code is publicly available.","authors":["Diyako Ghaderyan","Stefan Werner"],"cats":["cs.LG","cs.AI"],"primary":"cs.LG","published":"2026-11-12T14:38:34Z","topic":0},{"id":"2611.16856","title":"Sustainable Federated Fine-Tuning with LoRA: Environmental Impact and Performance Footprint","summary":"Federated Learning (FL) with Parameter-Efficient Fine-Tuning (PEFT) enables efficient adaptation of Large Language Models (LLMs), yet its energy and carbon costs across hardware remain poorly understood. We evaluated federated Low-Rank Adaptation (LoRA) fine-tuning on NVIDIA A40, RTX 6000 Ada, and A100 GPUs using a $2^2$ full factorial design with 135M/360M models and 10/100 local steps. CodeCarbon tracked energy and estimated CO$_2$-equivalent emissions. Descriptive allocation of variation assigned 84.8-91.2% of energy and emissions variation to local steps within these ranges. On A40 and RTX 6000 Ada, configurations with fewer steps achieved comparable accuracy with roughly 6-8-fold lower energy consumption and higher accuracy per kWh. These findings support smaller local budgets for energy-efficient federated fine-tuning within the evaluated setting.","authors":["Pedro Augusto Londe Ribeiro","Rodrigo Moreira","Larissa Ferreira Rodrigues Moreira","Joberto S. B. Martins","Tereza C. Carvalho","Flávio de Oliveira Silva"],"cats":["cs.LG","cs.NI"],"primary":"cs.LG","published":"2026-11-12T14:30:10Z","topic":4},{"id":"2611.16843","title":"Shortcuts or Memorisation? Revisiting Data Corruption in Natural Language Understanding with Large Language Models","summary":"Pre-trained encoders fine-tuned on natural language inference (NLI) and other language understanding tasks keep much of their accuracy after the input is perturbed, e.g. by removing all nouns or shuffling the words. Does the same hold for large language models (LLMs), and if so, is it because they have seen the benchmarks during pre-training? In this paper we revisit these perturbation experiments with eight LLMs alongside fine-tuned encoders on GLUE and the three rounds of ANLI. Regardless of model size, every system that answers directly retains 36-72% of its above-chance performance without nouns on MNLI compared to only 14-29% on ANLI. Only a model that reasons before answering departs from this pattern. We also check contamination indicators (including exact $n$-gram counts in the public pre-training data of two of the models) and find no link between exposure and robustness. Our strongest evidence comes from 368 new, human-validated items that none of the models can have seen. All eight LLMs keep more of their performance without nouns on items written in MNLI's style than on items written without its cues. When the perturbation instead removes the function words carrying such cues, this advantage reverses. In light of these results, we argue that robustness to perturbation on familiar NLI benchmarks reflects dataset shortcuts rather than memorisation, and tells us more about the benchmark than about the model.","authors":["Aarne Talman","Stergios Chatzikyriakidis"],"cats":["cs.CL"],"primary":"cs.CL","published":"2026-11-12T14:18:44Z","topic":6},{"id":"2611.16831","title":"A Modular Framework for ESG Prediction through Domain-Specific Language Modeling and Graph Learning","summary":"Environmental, Social and Governance (ESG) scores are increasingly used to evaluate corporate sustainability, supporting investment decisions, ensuring regulatory compliance and assessing risk. However, current ESG evaluations largely rely on proprietary methodologies that often produce inconsistent ratings. This motivates the development of data-driven prediction frameworks. In this paper, we present ESGPred, a modular framework for predicting ESG scores by combining domain-specific language modeling with graph-based learning over a network that models the relationships among the companies being evaluated. In addition to presenting the rationale behind ESGPred and detailing its ESG score prediction process, we show that ESGPred can effectively predict both overall ESG ratings and their individual Environmental, Social, and Governance components. We evaluated ESGPred using a dataset of 5,266 sustainability reports published by 2,134 publicly listed companies between 2020 and 2023. In this evaluation, we used ESG ratings provided by LSEG Data & Analytics as ground truth. Our experimental results demonstrated that integrating graph-based relational learning improves prediction performance over purely text-based baselines.","authors":["Sofia Baiocco","Gianluca Bonifazi","Christopher Buratti","Michele Marchetti","Camilla Mazzoli","Federica Parlapiano","Giulia Quaglieri","Davide Traini","Domenico Ursino","Luca Virgili"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T14:02:59Z","topic":6},{"id":"2611.16821","title":"Calibration and Conformal Prediction for Medical Question Answering in Arabic and English: Evidence from Two Non-Parallel Benchmarks","summary":"Large language models (LLMs) are increasingly tested on medical exam questions, but accuracy alone does not show when an answer can be trusted. I study the calibration and uncertainty of two open-weight models, Qwen2.5-7B-Instruct and Jais-adapted-7B-chat (both run in 4-bit precision), on 2,816 English multiple-choice questions from MedMCQA and 4,927 Arabic multiple-choice questions from MedAraBench. Accuracy was 0.561 (English) and 0.430 (Arabic) for Qwen, and 0.369 and 0.308 for Jais. Raw option probabilities were overconfident in all four settings (expected calibration error 0.242 to 0.486), and a single temperature parameter removed 82% to 93% of that error. Split conformal prediction reached its nominal coverage in all 24 settings I tested (within 0.006), but the sets were large when accuracy was low: at 90% coverage, temperature-scaled sets contained on average 63% (Qwen, English), 78% (Qwen, Arabic), 86% (Jais, English) and 89% (Jais, Arabic) of the options. The margin between the top two option probabilities separated correct from incorrect answers best for Qwen on English (AUROC 0.734) and worst for Jais on Arabic (0.579). Answering only the 20% most confident questions lowered Qwen's English error from 0.439 to 0.098. Jais showed a strong prior for option C (49% of English and 60% of Arabic predictions). Scoring Latin letters only cut the Arabic share to 29% but left accuracy, raw ECE and AUROC nearly unchanged. Because the two benchmarks differ in subject, level and number of options, I do not attribute differences between languages to language itself. Code is available at https://github.com/Mariyaaleem/arabic-english-medqa-uncertainty.","authors":["Mariya"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T13:57:23Z","topic":6},{"id":"2611.16816","title":"Outcome-Mass Alignment: Controlling Length-Dependent Trajectory Influence in Multi-Turn LLM Agent Reinforcement Learning","summary":"To keep long interaction contexts tractable, stepwise group-relative training for multi-turn LLM agents materializes each source trajectory into multiple response rows and valid tokens. This expansion creates an outer-accounting problem distinct from within-trajectory credit assignment: row-based outcome statistics can count one episode outcome multiple times, while token-level reduction can assign unequal aggregate objective mass across trajectories. We term these effects outcome-statistics multiplicity and objective-mass imbalance. We introduce Outcome-Mass Alignment (OMA), a two-component method that makes both accounting choices explicit. Outcome Alignment computes outcome statistics over unique source trajectories, while Mass Alignment controls aggregate objective mass after within-trajectory aggregation of optimizer-native token objectives. Neither redesigns the base optimizer's credit assignment, importance ratios, or clipping. Across matched ALFWorld GRPO comparisons at two model scales, where failed trajectories typically expand much more than successful ones, OMA improves matched-horizon endpoint success by 7.0-9.4 percentage points. Ordinary WebShop exhibits weak or reversed expansion-outcome coupling and correspondingly mixed effects, whereas in a success-oriented WebShop setting, WebShop-SOL, OMA improves success nAUC from 0.399 to 0.511. Component ablations and GiGPO and GSPO comparisons further support component separability and cross-optimizer compatibility. These results show that trajectory expansion always defines an implicit weighting rule, but its practical effect depends on how expansion couples to outcomes and optimizer-native updates.","authors":["Dawei Zhang","Yubin Huangfu","Jessica Fan","Boyu Wei","Likang Wu","Hongke Zhao","Ming He","Zhu Teng"],"cats":["cs.CL"],"primary":"cs.CL","published":"2026-11-12T13:52:12Z","topic":0},{"id":"2611.16803","title":"Detection Is Not Early Warning: Evaluating Token-Level Alarms on Language Model Failures","summary":"Many failures of language models begin at a specific token, the onset. A monitor that reads the model during generation can classify the finished response, detect the failure once it is under way, or warn before it begins, and only a warning leaves time to stop or steer the generation. Monitors are usually evaluated with response-level metrics such as AUROC, which ignore when the monitor fires. We introduce PREWARN, an evaluation protocol for any failure with an annotated onset. It calibrates each monitor's threshold on validation data to an error rate chosen for the application, runs the monitor, and scores its alarms relative to the onset on all three tasks. As a case study we apply it to repetitive degeneration in three open language models, with early warning as the main task, using linear probes on activations and six untrained baselines. Response-level metrics saturate on this failure, and a rule that reads only the response length scores almost perfectly, while probes that AUROC ranks as equal warn very differently under PREWARN. Probes warn before the onset on most degenerate responses of two of the three models, low-rank adapters below the probe more than double the share of tokens just before the onset that it flags, largely by flagging degenerate responses long before the onset, and no untrained baseline warns much more often than random alarms. We release the probes with calibrated thresholds at https://huggingface.co/collections/marcodene/prewarn-degeneration-probes","authors":["Marco De Negri","Luca Sartori","Lorenzo Baggi","Moritz Reihs","Eduard Frank Ďurech","Mennatallah El-Assady","Valentina Pyatkin","Anna Hedström"],"cats":["cs.LG","cs.CL"],"primary":"cs.LG","published":"2026-11-12T13:43:19Z","topic":6},{"id":"2611.16792","title":"Systematic Evaluation of Training Data Generation Strategies for Machine Learning-Based Molecular and Materials Property Prediction","summary":"Machine-learning (ML) models for molecular and materials property prediction are often trained on large datasets curated through arbitrary or generic sampling strategies. Such approaches can lead to the over- or under-representation of regions of chemical and materials space that are most relevant to the prediction task. An important challenge is therefore how to efficiently curate training data to maximize predictive performance while minimising the number of labelled examples required. Here, we systematically compare training data acquisition strategies for molecular and materials property prediction, ranging from random sampling to various active learning approaches. We evaluate their performance across different model classes, representations, datasets, and target properties. Our results show that the effectiveness of data acquisition strategies depend strongly on the interplay between the model, dataset, and prediction task, although uncertainty-based sampling generally outperforms alternative strategies. Based on these findings, we derive guidelines for selecting data acquisition strategies under a given labelling budget.","authors":["Minh Nghia Nguyen","Beate Paulus","Akhil S. Nair"],"cats":["cond-mat.mtrl-sci","cs.LG"],"primary":"cond-mat.mtrl-sci","published":"2026-11-12T13:38:38Z","topic":4},{"id":"2611.16789","title":"ScrewSense: Mapping the Invisible - Anomaly Detection and Error Classification in Screwdriving Processes","summary":"The growing volume of process data generated by screwdriving systems, combined with advancements in computing power, data storage, and algorithms, is facilitating the application of artificial intelligence in this field. Accurately detecting and classifying screwdriving errors is crucial for ensuring product quality, avoiding costly rework in manufacturing, and supporting sustainability by facilitating the maintenance, reuse, and recycling of screw-connected products. In this context, machine learning approaches offer a promising alternative to traditional monitoring techniques, such as window-based methods or envelope curves. This study examines the use of deep learning techniques for anomaly detection and error classification in screwdriving processes. The data corpus consists of multivariate time series capturing process parameters, such as torque and rotation angle. For anomaly detection, the embedding space of a vector quantized variational autoencoder was analyzed. In terms of error classification, four different model architectures were evaluated. The results demonstrate the effectiveness of anomaly detection and confirm that deep learning models can distinguish between multiple error categories beyond traditional binary classification. This extends the capabilities of conventional monitoring approaches, achieving accuracies of up to 100%. Additionally, real-world data material was examined in order to further demonstrate model robustness. Overall the results indicate strong generalization capability and highlight the practical potential of the developed models for use in industrial screwdriving processes, eliminating the need for expert knowledge in error identification and significantly reducing manual effort and analysis time.","authors":["Tobias Lettner","Tatyana Ivanovska","Anwar Al Assadi","Christian Bergler"],"cats":["cs.LG","cs.RO","eess.SP"],"primary":"cs.LG","published":"2026-11-12T13:35:41Z","topic":4},{"id":"2611.16772","title":"Class-Competence Aggregation: Output-Layer Routing for Federated Multiclass DDoS Detection under Class-Concentrated Non-IID Data","summary":"Federated learning over distributed intrusion-detection sensors naturally produces \\emph{class-concentrated} non-IID data, where an entire attack class is held by a single client and absent elsewhere. We show this collapses federated averaging in a specific, localized way: the output layer holds one decision boundary per class, and when a class's aggregation weight comes overwhelmingly from clients that have never seen it, its neuron receives an active anti-signal rather than mere dilution, collapsing to zero. On CIC-DDoS2019, standard FedAvg drives the concentrated LDAP class to $0.00 \\pm 0.00\\%$ F1 across every seed. We propose Class-Competence Aggregation (CCA), which aggregates the output layer per class using each client's own locally reported validation F1 as the weight, while hidden layers remain under standard size-weighted FedAvg. CCA requires no server-held auxiliary data and adds only one F1 scalar per class per round beyond the feature-scaling and class-count exchanges common to any federated pipeline. On our primary CNN-LSTM architecture, CCA alone raises LDAP F1 from $0\\%$ to $84.83 \\pm 1.65\\%$, and the full method reaches $91.64 \\pm 0.57\\%$ macro F1, closing the federated-to-centralized gap from $-42.53$ to $-4.05$ points. Benchmarked against four other federated methods under identical partitions and seeds, CCA is the only one that recovers the concentrated class substantially and reliably; a $178\\times$ smaller MLP shows the same qualitative recovery, though less completely, confirming the effect holds across architecture scale even as its magnitude varies. Results are validated across three seeds and replicated on a second dataset, CICIDS2017.","authors":["Tapu Datta","Hafiz Imtiaz"],"cats":["cs.LG","stat.ML"],"primary":"cs.LG","published":"2026-11-12T13:18:13Z","topic":7},{"id":"2611.16767","title":"Evaluating Video LLMs' Understanding of Corner Cases in Autonomous Driving","summary":"Natural language explanations promised by modern Video Large Language Models (Video LLMs) offer autonomous driving systems a path toward detailed scene description and interpretable decision making on the road. While recent benchmarks have begun assessing these models on driving scenes, they mainly target routine traffic conditions and general comprehension tasks based on existing driving datasets and therefore fail to evaluate model performance on the rarest and most safety-critical driving events, also known as \"corner cases\". To provide a comprehensive evaluation of Video LLMs on precisely these scenarios, we introduce DashBench, a benchmark testing 25 Video LLMs on 1,709 carefully curated driving videos spanning not only collision-related corner cases but also rare non-collision ones. Through extensive prompts and open-ended question answering, we expose a critical capability gap: all evaluated models perform substantially below the threshold required for safety-critical driving applications, frequently misjudging or missing dangerous events on the road ahead and hallucinating non-existent agents or actions in their explanations. We believe DashBench, combining an extensive dataset with a rigorous evaluation toolkit and a detailed diagnosis of current model limitations, drives progress toward safer, more explainable autonomous driving systems. The code and data will be shared at: cornercase-dashbench.github.io","authors":["Gabriel Manalu","Minh Son Hoang","Hyeontaek Hwang","Jaesung Lee","Hyojin Choi","Tran Tuan Phong","Seon Kwon Kim","Saad Wazir","Daeyoung Kim"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T13:15:27Z","topic":2},{"id":"2611.16756","title":"Beyond Binary Classification: Multiclass Machine Learning for Diabetes and Prediabetes Risk Prediction Using NHANES","summary":"Detecting diabetes risk early is important for preventing disease progression. Most existing machine learning approaches focus on binary classification, which limits their ability to capture intermediate risk states such as prediabetes. We develop a supervised learning framework to classify adults as normal, prediabetic, or diabetic using National Health and Nutrition Examination Survey data. The framework uses routinely collected demographic, clinical, and lifestyle variables from 15,356 adults, with glycated hemoglobin (HbA1c) used only to define labels and excluded from input features to prevent leakage. We compare conventional machine learning and transformer-based models and examine two binary formulations: normal versus prediabetes, and normal versus combined prediabetes and diabetes. Models are evaluated on a held-out test set using precision, sensitivity, specificity, and F1 score. In the multiclass setting, performance is broadly consistent across models, with accuracy up to 0.72 and macro-averaged F1 of 0.70. Prediabetes remains the most challenging class, with F1 scores of 0.54 to 0.56 for conventional models and 0.53 to 0.55 for transformers, reflecting its overlap with both neighboring states. When the task is reduced to normal versus prediabetes, F1 scores rise modestly to 0.62 to 0.64. A more substantial improvement is observed under a broader screening formulation combining prediabetes and diabetes, with F1 scores of 0.73 to 0.74 across all models. These findings suggest that feature overlap along the biological continuum between normal and diabetic states contributes more to the difficulty of prediabetes detection than model architecture. Multiclass prediction provides a more clinically informative representation of metabolic risk progression but highlights a trade-off between clinically meaningful stratification and predictive accuracy.","authors":["Aman Mujahid","Anita Penkova"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T13:06:18Z","topic":6},{"id":"2611.16752","title":"AgriSense: Vision-Language Decision Support and Robotic Intervention in a Virtual Greenhouse Copy title","summary":"Autonomous greenhouse management increasingly requires robotic systems that integrate perception, navigation, environmental regulation, and plant-level intervention. This paper presents AgriSense, a framework combining an Isaac Sim virtual greenhouse, a quadruped mobile manipulator, RGB and LiDAR-inertial sensing, environmental control, and supervised vision-language decision support. Frozen DINOv2 and BERT encoders, together with structured environmental inputs, support joint prediction of twelve environmental references, a target zone, and an intervention flag. Checkpoint replay reproduces 41/43 zone decisions. Across five seeds on a source-grouped test set, the full, image-removed, and language-removed configurations each achieve 43/44 correct zone decisions, providing no evidence of incremental multimodal zone-accuracy improvement. The full model achieves mean absolute errors of 0.249 degrees C for temperature, 2.030 percentage points for humidity, and 0.997 percentage points for soil moisture. Sensor-only inputs match full-model zone accuracy, while language-only inputs achieve 68.64%. External real-image benchmarks reveal distribution-shift limitations. Five offline Point-LIO routes yield rigidly aligned trajectory errors of 5.22-17.32 mm. Of 40 virtual-arm targets, 32 satisfy geometric acceptance criteria and eight are rejected at boundaries. Six long-horizon climate cases and fault protocols characterize integration limits. Navigation and integrated treatment rely on simulator-state inputs; physical greenhouse performance and agronomic treatment efficacy remain unvalidated. The contribution is an evaluated decision-support and robotic-intervention framework for a virtual greenhouse.","authors":["Lyes Saad Saoud","Oualid Doukhi","Reza Ghorbani","Moussa Ayyash"],"cats":["cs.RO","cs.AI","cs.CV"],"primary":"cs.RO","published":"2026-11-12T13:04:57Z","topic":2},{"id":"2611.16750","title":"Interpretable machine learning diagnosis of degradation mechanisms in lead-free $RbSn_{0.5}Ge_{0.5}I_{3}$ solar cells from temperature- and illumination-dependent electrical measurements","summary":"Machine learning coupled with device simulation is widely used to optimize thin-film solar cells, but almost exclusively in the forward direction, from device parameters to predicted performance. The inverse problem encountered after fabrication, namely identifying the mechanism that limits a given device, is still addressed by manual inspection of characterization data. In this work, both directions are examined for lead-free $RbSn_{0.5}Ge_{0.5}I_{3}$ solar cells using approximately 42,000 SCAPS-1D simulations. Replacing the Spiro-OMeTAD hole transport layer with CuI increases the simulated power conversion efficiency from 19.6\\% to 21.6\\%, almost entirely through the fill factor. Surrogate models trained and tested on a factorial parameter grid substantially overstate their predictive accuracy. An XGBoost regressor attains $R^2 = 0.9997$ on held-out grid points but only 0.58 on independently sampled devices, whereas training on a Latin hypercube design of comparable size restores $R^2 = 0.995$. For the inverse problem, a multi-label classifier identifies which of six degradation mechanisms are present from simulated temperature- and illumination-dependent open-circuit-voltage measurements. It achieves a macro $F_1$ score of 0.938, compared to 0.687 for a calibrated physics-rule baseline. The classifier does not, however, generalize to combinations of mechanisms absent from the training set.","authors":["Nabanita Sarker","Hafiz Imtiaz"],"cats":["cs.LG","math-ph"],"primary":"cs.LG","published":"2026-11-12T13:03:51Z","topic":1},{"id":"2611.16749","title":"ORCA:Orthogonal Residual Consensus Alignment for Multi-View Clustering","summary":"Deep multi-view clustering often suffers from redundant information accumulation during fusion and indiscriminate alignment across representation levels. We propose Orthogonal Residual Consensus Alignment (ORCA) to address these challenges. Its semantic incremental recurrent fusion module extracts cross-view orthogonal residuals as non-redundant increments and adaptively weights recurrent states through divergence-guided gating. Its consensus-calibrated hierarchical contrastive alignment module uses the global consensus to weight contrastive alignment of view-level, pairwise, and recurrent representations. Together, these modules balance view-specific structure with cross-view consistency. Experiments on five benchmarks against fifteen competing methods demonstrate ORCA's effectiveness, including a 5.25 percent accuracy improvement over the strongest baseline on 100leaves. Ablation studies validate both modules. Code is available at https://github.com/XiaojianDing/2026-NeurIPS-ORCA.","authors":["Xiaojian Ding","Xian Li","Xiaoying Zhu","Lin Zhao","Kaixiang Wang"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T13:00:39Z","topic":1},{"id":"2611.16748","title":"PhysLoop: Harnessing LLMs' Physics Awareness for Self-Improving Video Generation Models","summary":"Video generation models produce visually realistic videos but still struggle with physical dynamics and causal consistency. Learning from generated videos requires fine-grained physical supervision, yet reliable quality judgments, correction pairs, and controlled variations are difficult to obtain. Large language models (LLMs) possess physical knowledge, but struggle to reliably guide video generation models in producing their training videos. We present PhysLoop, a unified harness designed to support physically grounded self-improvement in video generation models. The harness guides and constrains LLMs throughout video generation, selection, and correction, distilling their physical knowledge into physically consistent videos. The framework supports controllable training-data construction and inference-time video repair. By identifying and correcting failures, PhysLoop marks a key milestone toward recursive self-improvement (RSI) in video generation models with limited human intervention. We release 100K videos spanning diverse, realistic indoor and outdoor scenes, along with generation tools and two evaluation datasets: PhysLoop-Gen and PhysLoop-Repair. We evaluate controllable generation and localized repair on these datasets, respectively, and assess physical behavior on the external Physics-IQ Verified benchmark.","authors":["Yuxuan Chen","Xiaojie Li","Hao Chen","Juncheng Ma","Kewei Zhang","Zhongjing Du","Guoqiang Wei","Fangfang Wang","Zhen Dong","Zhiyang Dou","Liming Zhao","Yan Zeng","Jianwei Zhang","Daquan Zhou"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T12:59:35Z","topic":3},{"id":"2611.16744","title":"Scale-Induced Semantic Shift in Visual Prompt Injection Against VLMs and Web Agents","summary":"Visual content in multimedia systems may be resized, compressed, or re-rendered before it is interpreted by a vision-language model. We study a security failure mode in which these transformations change the machine-readable semantics of an image: a textual instruction is not recovered at a $512\\times512$ presentation resolution under our evaluation protocol but becomes recoverable after aggressive downsampling. We formulate this behavior as Scale-Induced Semantic Shift (SISS) and develop a closed-loop black-box framework that combines dual-scale verification, selective self-hardening, shallow-layer low-rank adaptation, and pixel-space noise regularization. On a fixed 300-case benchmark, the third SISS refinement round obtains a $58.75\\%$ average attack success rate across four closed-source vision-language models, compared with $41.08\\%$ for the strongest generation baseline. A controlled sweep bounds the effect to aggressive compression: average success decreases to $17.68\\%$, $6.85\\%$, and $5.60\\%$ at $128\\times128$, $224\\times224$, and $336\\times336$, respectively. In a downstream web-agent study, the generated images increase hidden-instruction perception, but absolute behavioral compromise remains about $4\\%$. Generic input transformations provide little protection in a separate defense protocol, whereas the proposed ScaleGuard consistency detector reduces non-adaptive success from $61.4\\%$ to $24.3\\%$ and one-round defense-aware success from $42.5\\%$ to $15.0\\%$, with no observed rejection among matched clean controls. The results characterize a bounded but reproducible semantic inconsistency in multimedia preprocessing and motivate scale-consistent visual-language inference.","authors":["Yiheng Zhao","Runze Ji","Chenyu Yi","Yi Yu","Bingquan Shen","Alex Kot","Xudong Jiang"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T12:57:11Z","topic":0},{"id":"2611.16736","title":"On the Difficulty of Sequence-Level Supervised Learning for Autoregressive Language Models","summary":"During the supervised learning phase, autoregressive language models are typically trained by minimizing the negative log-likelihood of the training data, which decomposes as a sum of token-level cross-entropies via teacher forcing. This objective is simple and efficient but potentially suffers from so-called exposure bias: a mismatch between the way the model is trained and the way it is used at inference time. Can we instead design sequence-level loss functions that incorporate inference-time decoding procedures, such as greedy decoding or autoregressive sampling? We show that two natural attempts lead to loss functions that are fundamentally flawed: unbounded below. We further show that these two flawed loss functions are intimately connected: using autoregressive Gumbel tricks, the sampling based loss is the expected greedy loss under Gumbel perturbations. Since the pathology stems from the dependence of the decoded sequence on the model parameters, we analyze the update direction obtained by applying a stop gradient on the sampling procedure, leading to a moment matching direction. While it is not the gradient of any loss function, we prove that it defines a valid descent direction in function space for the autoregressive negative log-likelihood. Despite these theoretical insights, the moment matching direction suffers from high variance. We conclude that teacher forcing may well be ``all you need'' for supervised learning of autoregressive models.","authors":["Mathieu Blondel","Michaël E. Sander","Germain Vivier-Ardisson","Tianlin Liu","Vincent Roulet"],"cats":["cs.LG","stat.ML"],"primary":"cs.LG","published":"2026-11-12T12:51:08Z","topic":6},{"id":"2611.16728","title":"EgoX: Egocentric Cross-Embodiment Manipulation with Embodiment Dreaming","summary":"Generalist manipulation policies can benefit from the complementary experience of humans, real robots, and simulated agents. However, differences in morphology, sensing, and control interfaces make it difficult to learn effectively from their combined demonstrations. We introduce EgoX, an egocentric cross-embodiment learning framework that organizes heterogeneous demonstrations through semantically corresponding observation and action modalities. EgoX pretrains a Modularized Cross-Embodiment Transformer with Embodiment Dreaming (MXT+) on this mixture and fine-tunes it on target-embodiment demonstrations. Within each embodiment, MXT+ assigns a dedicated tokenizer to each observation modality and a dedicated expert to each action modality, with a Transformer trunk shared across embodiments. A training-only embodiment-dreaming objective predicts latent representations of future visual and proprioceptive states, encouraging the shared trunk to model each embodiment's dynamics. Across 13 task-embodiment pairs spanning five tasks and six robot embodiments, including five physical robots and one simulated humanoid, cross-embodiment pretraining raises MXT+'s average normalized task progress from 0.63 to 0.80 over training from scratch. The gains are particularly pronounced under out-of-distribution conditions, where average task progress increases from 0.53 to 0.77. Ablations demonstrate the benefits of broader pretraining mixtures, modular interfaces, and embodiment dreaming, with dreaming producing more similar latent geometry across embodiments. Project page: egox-dream.github.io.","authors":["Yaru Niu","Binghong Chen","Changwei Yao","Shangtao Li","Zhenlong Fang","Revanth Krishna Senthilkumaran","Shuai Zhou","Yuemin Mao","Kalie Ching","Po-Ju Chao","Yudong Luo","Peiqi Yu","Heng Zhang","Hao Zhang","Bingqing Chen","Chen Qiu","H. Eric Tseng","Ye Yuan","Umar Iqbal","Changliu Liu","Jonathan Francis","Ding Zhao"],"cats":["cs.RO","cs.AI","cs.LG"],"primary":"cs.RO","published":"2026-11-12T12:40:24Z","topic":2},{"id":"2611.16727","title":"From Foundation VLM to Versatile Driving VLM: A Triplet Incremental Training Approach","summary":"Adapting general Vision-Language Models (VLMs) to Autonomous Driving (AD) is thought to require heavy 3D spatial priors or intrusive architectural overhauls to compensate for a lack of spatial grounding. In this work, we present a counterintuitive pilot study: off-the-shelf general VLMs already exhibit remarkable adaptability across diverse driving tasks, while precise metric depth remains the primary missing link, a bottleneck readily addressable via onboard sensors and depth foundation models. To this end, we introduce the Triplet Incremental Training Framework, a non-intrusive paradigm that seamlessly instills 3D geometric awareness into a general VLM backbone while fully preserving its open-world capabilities. Driven by automatically constructed (image, depth, QA pair) triplets, our framework operates via a unified tri-stream training pipeline: an unfrozen primary VLM stream to adapt to driving tasks, an auxiliary depth stream that guides the visual backbone to implicitly imagine 3D scene geometry via explicit geometric constraints, and a frozen twin VLM stream employing logit-level KL distillation to prevent semantic drift. Crucially, the auxiliary depth branch is discarded during deployment, ensuring the enhanced geometric awareness comes with zero inference overhead. We evaluate our framework across four major benchmarks targeting four diverse capabilities spanning 3D spatial perception, driving-oriented language reasoning, end-to-end planning, and general multimodal understanding. Our approach outperforms state-of-the-art driving VLMs, rivals domain-specific expert models in 3D perception, and successfully retains its original general reasoning capabilities. This demonstrates that Triplet Incremental Training offers a unified, deployment-ready paradigm for transforming general foundation models into robust driving intelligences.","authors":["Shengtao Wu","Zichao Dong","Weixuan Qiao","Xiaolu Liu","Xin Zhan","Junbo Chen","Jianke Zhu"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T12:39:25Z","topic":2},{"id":"2611.16726","title":"From Data to Wisdom: Rethinking AI Through an Epistemic Lens","summary":"This paper proposes the Data-to-Wisdom Epistemic Architecture for AI (EAAI), a conceptual framework treating Data, Information, Knowledge, Understanding and Wisdom as five distinct but recursively interacting epistemic functions: observation, contextual interpretation, modeling and prediction, critical evaluation and informed judgment. Two observations motivate it: sequential readings of the Knowledge Pyramid obscure how these functions interact, and a targeted review of ten Bayesian reference books finds that prior distributions are defined mathematically while \"prior information\" and \"prior knowledge\" are not consistently distinguished as epistemic categories. A Bayesian interpretation relates the likelihood, prior and posterior distributions, posterior prediction, model checking and decision analysis to the five functions, without limiting EAAI to Bayesian systems. Extending \"prior\" in an epistemic sense, the paper distinguishes five kinds of prior context that can shape the model, the likelihood, the prior, evaluation and action. Three worked applications -- quantitative financial risk, actuarial modeling in health insurance and agentic AI for medical reimbursement claims -- show how EAAI structures the analysis of model assumptions, evidence interpretation, validation, dependencies and decision consequences, separating the prediction of a risk, the appraisal of its reliability and the justification of action. It draws implications for AI and machine learning, including evaluation by function, treating intermediate model outputs as evidence open to challenge and separating generation from appraisal, and compares EAAI with work on DIKW, Bayesian workflow, AI evaluation and risk governance. Its contributions are an epistemic vocabulary with recursive relationships, a more accessible reading of Bayesian methodology, a typology of prior context and clearer methodological terminology.","authors":["Afonso Eduardo"],"cats":["cs.LG","cs.AI"],"primary":"cs.LG","published":"2026-11-12T12:38:36Z","topic":0},{"id":"2611.16722","title":"Mind the Accuracy-Approval Gap: How Users Evaluate Sycophantic AI in Everyday Tasks","summary":"Large language models (LLMs) often exhibit sycophancy, a tendency to agree with users, flatter or affirm them, and preserve their face, even when independent judgment or critical feedback would be more useful. Prior work has largely evaluated sycophancy through benchmark evaluations, but less is known about how users perceive and respond to sycophantic AI in everyday tasks. We combine benchmark evaluations with two large-scale user studies. We study four personas defined through system prompts: objective (instructed to prioritize accuracy over agreement), soft-sycophantic, hard-sycophantic, and baseline (no persona-specific system prompt). Benchmarks show that even baseline exhibits substantial sycophancy, agreeing with users' stated views 71.7% of the time on NLP Survey, compared with 54.8% for objective and 89.5% for hard-sycophantic. Under conversational pressure on TruthfulQA, accuracy ranged from 81.9% for objective to 59.6% for baseline and 8.7% for hard-sycophantic. Participants were assigned to one of the four personas, and either discussed contested policy topics or collaborated with AI on email writing. Despite its higher accuracy under pressure, objective AI was rated significantly less respectful and satisfying than the other personas, and less trustworthy than baseline. Baseline and sycophantic personas did not differ significantly. Persona did not significantly affect how often participants changed their opinions, although participants who discussed with objective AI had greater odds of reduced confidence. Objective AI prompted greater writing expansion than soft-sycophantic AI, but final writing quality did not differ significantly across personas. These findings highlight a tension between accuracy and user approval: personas with substantially lower accuracy under conversational pressure received more favorable ratings for respect and satisfaction than objective AI.","authors":["Wajiha Naveed","Brandon Fain"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T12:35:03Z","topic":6},{"id":"2611.16718","title":"MusicRWKV: A Multimodal RWKV Framework for Music Representation Learning","summary":"Recent music foundation models have achieved remarkable progress in music representation learning, yet most existing approaches rely on Transformer architectures whose quadratic attention complexity introduces substantial computational and memory overhead when modeling long-duration music recordings. In this work, we propose MusicRWKV, a multimodal music representation learning framework built upon the RWKV architecture. MusicRWKV jointly models waveform, spectrogram, and metadata information within a unified self-supervised learning framework. To improve efficiency for long musical sequences, we introduce an invertible compression module that reduces sequence length while preserving reconstruction capability. A FiLM-based conditioning strategy is employed to facilitate modality-aware representation learning without introducing expensive cross-attention operations. We further investigate the integration of Soft ROSA as a differentiable retrieval mechanism for neural-symbolic interaction. Experimental evaluation across a diverse collection of Music Information Retrieval (MIR) benchmarks demonstrates that the proposed RWKV-based architecture can be effectively adapted to heterogeneous music understanding tasks. A controlled pretraining ablation shows that Soft ROSA improves validation performance, while layer-wise telemetry analysis reveals a progressive transition from distributed retrieval to selective retrieval in deeper network layers. These results suggest that RWKV provides a promising architectural alternative to Transformer-based models for multimodal music understanding.","authors":["Terry Tan"],"cats":["cs.SD","cs.AI","eess.AS"],"primary":"cs.SD","published":"2026-11-12T12:34:08Z","topic":3},{"id":"2611.16714","title":"Towards Green Text Generation: Memory-Based Speculative Decoding","summary":"Speculative decoding accelerates text generation with large language models by letting a cheap draft model propose tokens that a large verifier checks in a single forward pass. We introduce Olifant, a draft model built on a memory-based classifier. It predicts the next token from a four-token window in under 0.3 ms on CPU, requiring no GPU, no backpropagation, and no accelerator memory. We measure throughput and energy under two verifier regimes on consumer hardware. Without a KV cache, drafting is nearly free: a domain-matched Olifant model reaches 38.1% acceptance on EU legal text, delivering 2.43x throughput and 2.51x lower energy per token than the neural verifier model alone. With a KV-cached verifier, itself a ~10x energy saver, drafting costs matter: a TinyLlama draft model halves throughput and triples energy despite high acceptance, whereas Olifant still yields 1.25x throughput and the lowest energy in the study (93 J per 100 tokens, 14x below the uncached baseline). We further show that KV caching decouples time from energy: verification of longer drafts raises power draw, shifting the energy-optimal draft length below the speed-optimal one. Speculative decoding is not dead in the age of fast verifiers, provided drafting is nearly free.","authors":["Antal van den Bosch"],"cats":["cs.CL"],"primary":"cs.CL","published":"2026-11-12T12:32:14Z","topic":4},{"id":"2611.16710","title":"Acoustic Respiratory Phase Detection on Mobile Devices","summary":"Breathing monitoring is important for well-being, stress regulation, and guided breathing exercises, yet existing approaches often rely on contact-based sensors or controlled environments. This paper presents a non-invasive framework for respiratory phase detection using acoustic signals recorded on commodity smartphones. The proposed hybrid CNN-Transformer architecture combines local spectral feature extraction with long-range temporal modeling to capture the cyclic structure of breathing. The model is evaluated using a Leave-One-Subject-Out (LOSO) protocol on the Foundation Dataset and on an independent unseen test set comprising over 1300 breathing cycles and 2 hours of continuous recordings. The proposed 3-class formulation achieves 72.85% frame-level accuracy and 96.6% cycle-level exhalation detection efficiency in the LOSO evaluation, while achieving 76.94% frame-level accuracy, a Cohen's Kappa score of 0.65, and 99.2% cycle-level exhalation detection efficiency on the unseen test set. Results suggest that accurate respiratory monitoring may be achievable under data-constrained conditions, indicating that appropriate temporal modeling can learn transferable respiratory representations and generalize to previously unseen participants. An end-to-end ONNX deployment pipeline enables privacy-preserving on-device inference with latency below 20 ms, supporting real-time monitoring on mobile and wearable devices.","authors":["Julian Szymanski","Tomasz Sankowski","Piotr Sulewski","Iwo Czartowski","Javier Medina Quero"],"cats":["cs.AI","cs.LG"],"primary":"cs.AI","published":"2026-11-12T12:26:15Z","topic":1},{"id":"2611.16704","title":"Forecasting Sales Delay through Time-aware Analysis of Sales Events in Textual Reports","summary":"Business-to-business (B2B) companies develop and deliver a one-of-a-kind product in response to a client's orders. To prepare resources for development beforehand, companies estimate dates when they will receive orders from clients in their sales process, called scheduled order dates. However, dates when the companies actually receive orders can differ from the scheduled order dates. This difference increases costs in the development. If the difference could be forecasted before receiving orders, companies could optimize their resource preparation. Towards reducing the costs, we introduce sales delay forecasting, a novel task of forecasting delay days from the scheduled order dates during sales process where sales representatives (sales reps) interact with the clients to receive orders. Sales delay forecasting is challenging because the sales process progress inherently involves uncertainty and depends on the client's willingness to order. To address this task, we propose confidence features which represent sales reps' confidence as a proxy measure of the client's willingness to order. To represent it, our method extracts dates of future sales events anticipated by sales reps from textual reports. To consider whether the confidence objectively looks reasonable, we also introduce status features which describe sales process status based on current date and past events. In experiments using real sales data, our method achieved a 16% lower mean absolute error compared with a feature-based baseline similar to a prior work. Moreover, our method outperformed non-feature-based baselines using large language models, indicating that it is crucial to explicitly create our features.","authors":["Masaya Tsunokake","Yuta Koreeda","Yuhei Niwa","Terufumi Morishita","Teppei Inoue","Hiroaki Ozaki","Yasuhiro Sogawa"],"cats":["cs.CL"],"primary":"cs.CL","published":"2026-11-12T12:21:09Z","topic":6},{"id":"2611.16694","title":"When Expert Disagreement Hurts: Prestige and Selectivity in LLM Decision Revision","summary":"When language models revise a decision after another agent disagrees, neither reversal counts nor final accuracy shows whether they are fixing mistakes or abandoning correct answers. We introduce a controlled two-pass audit that holds the question, the evidence, and the model's initial answer fixed and varies only the revision context: neutral reflection, anonymous disagreement, or disagreement attributed to a domain expert. For binary decisions the accounting is exact: revision raises accuracy only if it is selective enough, reversing wrong answers more than $a/(1-a)$ times as often as correct ones, where $a$ is initial accuracy (nine times at 90%). Across five models and 1,989 evidence-grounded decisions in medicine, science, and law, no model clears this bar under expert disagreement, and accuracy falls by up to 16.1 points. The expert label, a prestige cue, adds reversals for four of the five models, yet its effect on accuracy ranges from a 5.2-point loss to a 2.4-point gain, so reversal counts are a poor proxy for harm. With identical arguments from both sources, the label still raises reversal for three proprietary models and in a pooled analysis of four newer models. Models also respond to stated source reliability. Revision pipelines should therefore be evaluated with paired, conditional metrics, per model and protocol.","authors":["Yupeng Tang","Mingfeng Lin"],"cats":["cs.AI","cs.CL"],"primary":"cs.AI","published":"2026-11-12T12:11:03Z","topic":6},{"id":"2611.16693","title":"MobileProbe: Adaptive Security Evaluation of Mobile Agents via Trajectory-Aware Injection","summary":"Mobile GUI agents read the screen of a phone and operate apps on a user's behalf to complete multi-step tasks such as online shopping or messaging. Because the agent acts on whatever the screen shows, an attacker can embed malicious instructions into ordinary UI content and redirect its actions. We refer to this threat as environmental injection attack (EIA). Existing benchmarks measure this risk by pinning hand-crafted payloads to a fixed step of a fixed task. This approach faces two limitations. First, each payload is bound to one task in one app and does not transfer without manual redesign. Second, agents given the same task may follow different routes, so an agent that never reaches the injected page is recorded as safe without being tested. This insufficient test coverage can cause fixed-point benchmarks to underestimate the security risk of deployed mobile agents. In this work, we propose MobileProbe, a trajectory-aware red-teaming framework that evaluates mobile agents adaptively. MobileProbe discovers attacker-controllable content surfaces through automated pentesting, localizes vulnerable decision points along the agent's trajectory, and iteratively refines context-aware payloads through a surrogate-filtered evolutionary loop. We construct a benchmark of 83 long-horizon attack instances spanning 12 mobile applications. Each is annotated with a Confidentiality, Integrity, or Availability (CIA) attack goal to cover the principal categories of security threats facing mobile agents. On three representative agent architectures, MobileProbe attains higher attack success rates than human-written, zero-shot LLM, and prior automated red-teaming baselines adapted to the mobile domain. MobileProbe can serve as a pre-deployment auditing tool to systematically uncover vulnerabilities and inform security hardening of mobile agents.","authors":["Yingbin Jin","Ji'an Wang","HaoYang Chai","Hanjun Luo","Li Hu","Xinfeng Li","Huadi Zheng","PeiZhao Hu","Qingqing Ye","Haibo Hu"],"cats":["cs.CR","cs.AI"],"primary":"cs.CR","published":"2026-11-12T12:08:43Z","topic":0},{"id":"2611.16678","title":"Edge-Gradient Augmentation under Reinforcement Learning for Temporal Graph","summary":"Temporal link prediction is critical for understand- ing dynamic interactions in evolving networks, informing appli- cations ranging from social network analysis to recommendation systems. Existing methods typically rely on raw temporal graph data, often neglecting explicit edge-change information that can significantly improve predictive performance. To address this gap, we propose edge-gradient reinforced learning (EG-RL), a framework that explicitly encodes edge-gradient events (edge births and deaths) through dedicated node and link encoding modules. EG-RL uses a policy-driven reinforcement-learning mechanism to dynamically select and augment the most infor- mative edge snapshots, guided by critic-based feedback to opti- mize prediction accuracy. The model comprises distinct modules for node encoding, link encoding, token mixing, and channel mixing, integrated through global pooling and followed by a two-layer multilayer perceptron classifier. Extensive experiments on multiple dynamic graph datasets demonstrate substantial gains in area under the receiver operating characteristic curve and average precision. Ablation studies further confirm the necessity of adaptive edge-gradient augmentation and validate the effectiveness of the proposed encoding and reinforcement- learning mechanisms.","authors":["Yongqing Zhou","Jiawei Chen","Wenjun Wang","Xinyu Wang","Wei Yu","Qiyao Peng"],"cats":["cs.LG","cs.AI"],"primary":"cs.LG","published":"2026-11-12T11:53:35Z","topic":2},{"id":"2611.16677","title":"Manifold Attracting Flow Matching: Physics-Constrained Sampling via Continuous Geometric Dynamics","summary":"Generating scientific data under previously unseen equality constraints requires satisfying the prescribed constraints without destroying the physical structure learned from data. Existing inference-time approaches often enforce physical constraints by repeatedly correcting predicted endpoints, but such discrete corrections can distort learned generative structure and introduce unstable feedback. In this work, we propose Manifold Attracting Flow Matching (MAFM), which continuously evolves a persistent anchor through an ODE whose dynamics drive it analytically toward the constraint manifold. Specifically, its geometric correction preserves model-guided tangent motion while prescribing the normal motion to contract constraint violations toward the manifold. This construction separates model tracking from constraint attraction and provides theoretical guarantees for residual contraction and local feedback stabilization. Experiments across geometric and partial differential equation benchmarks demonstrate improved generation quality over variant constraint-guided baselines while maintaining low constraint residuals. MAFM provides a flexible framework for coordinating learned physical structure and explicit constraints through controllable sampling dynamics, without retraining the underlying generative model.","authors":["Guanyu Chen","Jinchen Wang","Shengze Xu","Xingyu Ren","Tieyong Zeng","Antonios Armaou","Dong Ni"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T11:53:28Z","topic":7},{"id":"2611.16666","title":"Natural Gradient Learning in Predictive Coding through Lateral Circuits","summary":"Predictive coding (PC) trains neural networks through local energy-minimization dynamics without depending on a global error signal that backpropagation (BP) requires. Although PC is a promising alternative algorithm for efficient learning, the computational cost of inference dynamics currently outweighs the resulting learning benefits in practice. In this work, we take inspiration from optimization theory to extract more learning benefit from each inference phase. We introduce Natural PC, which augments standard PC with a lateral circuit, enabling exact and efficient natural gradient learning locally in space and time. Our lateral circuit learns the activity correlations that determine the Fisher information matrix and implicitly computes its inverse, avoiding expensive matrix inversion. Across all tested image reconstruction benchmarks, Natural PC achieves state of the art in terms of sample and compute efficiency compared to existing PC approaches. Most notably, Natural PC scales to ImageNet and matches the performance of BP trained with AdamW and various second-order optimizers at an order of magnitude lower computational cost while also using two orders of magnitude fewer training samples. Our results demonstrate that local natural gradient learning makes PC competitive with BP in both sample efficiency and computational cost in reconstruction tasks.","authors":["Lhea Beumer","Sander de Haan","Niclas Pokel","Pau Vilimelis Aceituno","Benjamin F. Grewe"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T11:42:17Z","topic":3},{"id":"2611.16662","title":"MuSeToM: Multi-Sensory Evaluation of Belief Revision with Textual Cues Across Six Senses","summary":"Theory of Mind (ToM), the ability to infer mental states in others, is pivotal for human social cognition. Existing evaluations of ToM in large language models (LLMs) mainly focus on static belief attribution and false-belief reasoning, leaving a critical aspect of social reasoning underexplored: can LLMs decide whether perceptual evidence is sufficient to justify another person's belief revision? To address this gap, we present MuSeToM, a human-validated, text-based benchmark for evaluating epistemic belief revision in ToM reasoning. MuSeToM contains 2,415 instances adapted from BigToM and OpenToM, with textual cues spanning six sensory categories, and tests whether models can determine when a protagonist should revise a belief and infer the resulting updated belief state. Using MuSeToM, we systematically evaluate state-of-the-art LLMs under direct and chain-of-thought prompting, revealing a pronounced dissonance: while models often detect that a belief may change, they struggle to identify the updated belief and frequently over-update on perceptually salient but epistemically insufficient cues.","authors":["Chunkit Chan","Kawai Chung","Sze Heng Douglas Kwok","Yuxuan Liu","Yauwai Yim","Tianshi Zheng","Haochen Shi","Qing Zong","Wong Kai Chung","Pak Hin Lau","Calvin Yang","Yifan Gao","Xiaoman Pan","Ginny Wong","Simon See","Yangqiu Song"],"cats":["cs.CL","cs.AI"],"primary":"cs.CL","published":"2026-11-12T11:38:52Z","topic":5},{"id":"2611.16661","title":"Reducing Batch Makespan for LLM Agent Workloads through Inference-Environment Coordination","summary":"Efficient batch execution of agent tasks is critical to LLM training and development workflows, including benchmark evaluation, synthetic-data generation, and RL rollouts. Since these tasks alternate between LLM inference and environment interaction, we present coordinated optimizations to reduce batch makespan, the time required to complete all tasks in a batch. For hybrid LLMs, a workload-based resource model guides GPU-memory allocation between key-value (KV) cache and recurrent state and provides a reference sample concurrency to exploit parallelism while limiting recomputation. Trajectory-aware routing and scheduling promote cache reuse while balancing worker load. Under constrained environment capacity, we decouple sample admission from environment execution and introduce Chain-Advancing Action Scheduling (CAAS), which prioritizes long chains using observed progress to shorten the batch tail. Controlled replay combines actual inference with recorded environment responses and durations for ablations of individual optimizations. Live evaluations with Ling-3.0-Flash and Ling-3.1-Flash on the BFCL, SWE-bench Multilingual, WideSearch, and MCP-Atlas benchmarks demonstrate makespan reductions of 10.4% to 65.0%, while evaluation scores remain aligned. Furthermore, these methods have been successfully applied to training and evaluation workflows for the Ling-3.x model series, accelerating the model-development cycle.","authors":["Wei Feng","Meiling Chen","Boxiao Chen","Lu Liu","Bin Zi","Libin Ye","Xingyu Lu","Lei Xu","Jun Zhou"],"cats":["cs.DC","cs.AI"],"primary":"cs.DC","published":"2026-11-12T11:38:21Z","topic":0},{"id":"2611.16657","title":"Price co-movement and price information diverge across Nordic electricity zones after the energy crisis","summary":"After the 2021-2022 energy crisis, day-ahead prices in Nordic bidding zones no longer move together. We ask whether this divergence also changes the information carried by prices. The answer matters for integrating variable renewable energy (VRE), because it can help identify where wind and solar forecasts are worth producing, publishing, and using. We measured the co-movement of each Norwegian and Swedish zone with Germany-Luxembourg (DE-LU) in 2018-2025. We then used a single model, selected on 2024 data, to forecast 2025 prices at 11:00 on the previous day and tested which inputs improved accuracy in each zone. The zones moved alike until 2019, but co-movement diverged in Sweden in 2020 and in Norway in 2021, and the split has persisted. In 2024, four southern zones co-moved strongly with DE-LU (intraday correlations of 0.54-0.71), while four northern zones co-moved weakly (0.25-0.30). Yet price information did not follow this split. Previous-day neighbouring prices reduced mean absolute error by at most 2.2% in any zone, equally in coupled and isolated zones. Removing the transmission system operators' wind and solar forecasts increased mean absolute error by at most 6% in Norway, 8-16% in Sweden, and 31% in DE-LU, and the ordering held in 2024. Within Norway, the value of these forecasts showed no detectable dependence on reservoir filling. The divergence in co-movement was therefore not matched by a divergence in price information, which differed mainly between national generation systems. Thus, a zone's coupling to the continent does not indicate the value of VRE forecasts there.","authors":["My Thi Diem Phan","Tuyen Trung Truong","Hoai Phuong Ha","Dat Thanh Nguyen"],"cats":["econ.GN","cs.LG","stat.AP"],"primary":"econ.GN","published":"2026-11-12T11:33:10Z","topic":7},{"id":"2611.16648","title":"GNN-Enhanced TCN Algorithms for ECG Anomaly Detection","summary":"To a time series can be also given the structure of directed graph, opening the doors to the usage of graph neural networks (GNNs) in this context. In this paper we develop a new pipeline involving novel geometric deep learning models that merge the capabilities of ordinary TCNs with the ones of GNNs to solve an anomaly detection problem on electrocardiogram signals.","authors":["Angelica Simonetti","Ferdinando Zanchetta"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T11:24:41Z","topic":7},{"id":"2611.16647","title":"InertiaMPDD: Multimodal Personalized Depression Detection via Emotional Inertia-Inspired Dynamics Modeling","summary":"Multimodal depression screening estimates risk from speech, visual behavior and language. Current methods mainly predict labels, with limited evidence of departures from a non-depressed reference. Recording-level supervision also leaves fine-grained emotional dynamics unannotated. Motivated by emotional inertia, the persistence of affect, InertiaMPDD learns healthy next-state patterns using a world-model-inspired latent predictor. Using pretrained multimodal representations, it predicts from observed history and context. Reference deviation, residual decay rate and a Jacobian spectral descriptor form reference-derived emotional dynamics descriptors, evaluated as candidate proxy biomarkers for standalone screening and personalized decision correction. Across five cohorts from four datasets, fusion reaches 98.41% F1 on D-Vlog and 84.98% on LMVD. Standalone descriptors outperform fixed additive fusion on MPDD but discriminate weakly on AVEC2014. These findings distinguish reference informativeness from fusion benefit and reveal context-dependent utility.","authors":["Changzeng Fu","Songyang Wang","Chang Gao","Xinyi Zhang","Peilin Li","Mo Meng","Jiarui Wu","Mingze Li","Hongwei Ding","Shiqi Zhao","Yong Li","Lang He"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T11:22:19Z","topic":3},{"id":"2611.16646","title":"Measurement-Calibrated Synthetic Pretraining for Rapid Adaptation of UAV Channel Predictors","summary":"Unmanned aerial vehicle (UAV) links can change faster than route-specific channel data can be collected, so a channel predictor must often be adapted from only a short target-flight calibration segment. This paper studies measurement-calibrated synthetic pretraining as an offline source prior for rapid adaptation of channel state information (CSI) predictors. The synthetic source is calibrated to flight-only power evolution, temporal shape variation, frequency selectivity, and phase disturbance, and is then combined with measured source-flight fine-tuning before target-flight adaptation. Under chronological held-out flight splits, synthetic-to-real (S2R) fine-tuning reduces the measured-data budget needed to reach useful prediction in the low-data regime. For 50 ms-ahead prediction, it crosses a latest-channel-estimate persistence baseline with 50 measured training windows, roughly 0.5 s of 10 ms-spaced supervised windows, whereas training from scratch requires 600 windows. A target-prefix adaptation study further shows that a lightweight gated recurrent unit predictor can reuse the pretrained source prior with one prediction-layer update epoch, positioning measurement-calibrated synthetic pretraining as a reusable prior for scarce target-flight measurements.","authors":["Qiuheng Zhou","Wei Jiang","Donglin Wang","Hans D. Schotten"],"cats":["eess.SP","cs.LG"],"primary":"eess.SP","published":"2026-11-12T11:19:06Z","topic":7},{"id":"2611.16645","title":"Rethinking Actions: Continuity-Aware Priors for World-Model Planning","summary":"Actions in expert demonstrations exhibit temporal structure. An ordered-versus-shuffled analysis across five environments confirms this dependence. We ask whether action proposals that better reflect expert continuity can improve visual world-model planning. Rather than imposing generic smoothness, we learn continuations from the temporal structure of demonstrated actions. Our locality-gated deep-kernel Gaussian-process prior, LG-DKGP, combines recent and longer-range executed action history, predicts a neural continuation, and learns a temporal Gaussian-process (GP) residual. A training-derived exploration scale initializes search around the corrected mean; a frozen world model evaluates progress toward the goal. Training first fits the neural mean, then jointly optimizes the prior with the GP residual using expert future actions. The resulting planner achieves 87.07\\% macro success across five simulated tasks spanning native and fragment-level goal attainment, compared with 71.03\\% for INTACT, 61.66\\% for PRISM, and 58.66\\% for LeWM. Component interventions and expert-relative action-structure comparisons support demonstration-informed continuations and reveal task-dependent effects of visual conditioning, history length, and planning horizon.","authors":["Shaobo Yang","Zichen Ai","Ziyang Ma","Qian Yang","Bin Fang"],"cats":["cs.RO","cs.LG"],"primary":"cs.RO","published":"2026-11-12T11:18:50Z","topic":5},{"id":"2611.16642","title":"Dataset Distillation via Drifting","summary":"The task of dataset distillation concerns the learning of a small set of synthetic samples that can serve as an effective replacement for a large real dataset when used for downstream tasks. While many solutions have been proposed since the problem's inception, most of them primarily seek to optimize the distilled data with respect to a classifier. In this work, we propose a new distillation method that learns a set of samples such that they directly approximate the per-class data distribution. Specifically, we \"drift\" our distilled data in the latent space of a feature extractor such that they move both toward the real data and away from each other without using a classification objective. We find that our drifted data cover the real distribution significantly better than existing methods while still serving as powerful training samples for classification. Furthermore, we showcase our method's use as a visualization tool for better interpreting models and extend it beyond images to other modalities. Our work advances the frontier of dataset distillation, offering a novel synthesis method and proposing distribution-based metrics as a complementary evaluation. More broadly, our findings help address the growing need to understand both data and models by uncovering more about how they interact.","authors":["George Cazenavette","Angelina Quan","Giannis Daras","Antonio Torralba","Vincent Sitzmann"],"cats":["cs.LG","cs.AI","cs.CV"],"primary":"cs.LG","published":"2026-11-12T11:17:40Z","topic":4},{"id":"2611.16639","title":"Asymmetric Multimodal Learning with Decomposed JEPAs","summary":"Cross-modal alignment is the standard objective for building multimodal representations. We show that alignment as a \\emph{sole} objective is lossy wherever modalities are informationally asymmetric---where neither can express what the other carries. Visual question answering is an obvious case: a question is a selective relational operator probing a dense visual field whose fine-grained features no query recovers, so what alignment preserves is not what the task requires. We prove an \\textbf{Alignment Redundancy Bound}: alignment-only Dirichlet energy minimization retains at most the redundant information atom $\\Red(Z; X_1, X_2)$, annihilating private modality-specific structure $\\Unq(Z; X_m)$ and cross-modal synergy $\\Syn(Z; X_1, X_2)$. We introduce \\textbf{Decomposed JEPA (D-JEPA)}, which factorizes latents into an aligned shared core and private modality-specific fibers. The shared core retains most of the structure alignment targets; the private fibers recover what alignment discards, via contextual masked prediction and asymmetric cross-subspace independence. Under information asymmetry, alignment-only training undergoes catastrophic semantic collapse on compositional reasoning benchmarks, surviving only on binary queries. D-JEPA averts this collapse, substantially improving downstream accuracy over both alignment-only and alignment-with-reconstruction baselines, and restoring a positive multimodal synergy gap in place of the baseline's inverted gap. Information-theoretic audits confirm that D-JEPA recovers most of the mutual information available from a frozen backbone, and causal fiber-swap interventions demonstrate that the recovered information is causally localized in the private fibers.","authors":["Yusuf Shihata"],"cats":["cs.LG","cs.CV"],"primary":"cs.LG","published":"2026-11-12T11:13:13Z","topic":3},{"id":"2611.16634","title":"Machine Learning for Scalable Multi-AP Coordination in IEEE 802.11bn Networks","summary":"As Wi-Fi deployments become denser, fully distributed random channel access no longer guarantees the required quality of service. The multi-access point coordination (MAPC) framework introduced in IEEE~802.11bn lets neighboring access points (APs) cooperate, but choosing which APs transmit together and how to configure them remains an open, computationally hard problem. We develop and validate a stack of machine learning (ML) coordination methods, progressing from lightweight hierarchical multi-armed bandits (H-MABs) to a generative pipeline built on graph neural networks (GNNs) and flow matching (FM). The methods are evaluated in simulation and on a hardware testbed, improving throughput over legacy access and scaling coordination to dozens of APs where analytical and search-based methods become intractable.","authors":["Maksymilian Wojnar","Szymon Szott"],"cats":["cs.NI","cs.LG"],"primary":"cs.NI","published":"2026-11-12T11:07:02Z","topic":2},{"id":"2611.16630","title":"When Chekhov's Gun Fires Anyway: Over-Realization in Text-to-Image Models","summary":"In text-to-image generation, a target that the prompt mentions often appears in the generated image in a way the prompt does not call for. We call this failure over-realization. Mental-spaces theory points to the situations where it can happen: under a negation, inside what a character believes or imagines, as the vehicle of a figurative comparison, or behind a viewpoint that hides the target. Prior work has studied only negation in depth, and the conventional alignment and preference metrics do not penalize the failure. We build OverReal to measure it: OverReal-Gen, 698 prompts covering the four situations, on which we measure each generator's over-realization rate; and OverReal-Det, 2,841 generated images human-labeled for whether and how the target is over-realized, on which we verify the automatic judge that produces this rate. Our evaluation of eight text-to-image systems reveals that all of them over-realize, and that conventional metrics such as CLIPScore reward the failure rather than penalize it. Prompt expansion only partly mitigates the failure. Inside the generator, the diffusion backbone attends to a mentioned target almost as much as to a requested one. The cue words that should hold the target back receive less attention than ordinary words.","authors":["Jiahao Huang","Kon Woo Kim","Song-Duo Ma","Yuka Kitamura","Felix Wagner","Yunshan Huang","Sunisth Kumar","Xanh Ho","Yun-Ang Wu","Zhijie Chen","Yunjia Xu","Junfeng Jiang","Zefan Yu","Fei Cheng","Akiko Aizawa"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T11:03:25Z","topic":3},{"id":"2611.16628","title":"Safety as a Budget: Calibrating the Control Channel of Constrained GRPO for Vision-Language Models","summary":"Reinforcement learning can teach vision-language models (VLMs) to decline harmful image-text requests, but how safe the result becomes is set indirectly, through reward weights found by trial runs. Stating safety as budgets on the rate of harmful assistance and the rate of over-refusal, enforced by Lagrange multipliers inside GRPO with an off-the-shelf scorer as the cost, is the natural fix, and it fails twice. The multiplier compares the scorer's mean with the budget, and scorers built for ranking get the mean wrong: a fixed-point analysis shows that the mean bias, not the budget, then decides whether the multiplier settles, saturates or switches off. And standard GRPO standardizes the penalized reward within each group, which cancels the multiplier before it reaches the gradient. Two-channel constrained GRPO repairs both: the base reward keeps the raw score for the within-group comparison, the penalty and the constraint estimate pass through a monotone map fitted on judge-labeled training responses, and the advantage keeps the multiplier. On a Qwen3-VL-2B policy with a training-free readout of an open 4B model as the scorer, held-out harm on 1,037 adversarial prompts falls from 8.9% under standard GRPO to 4.8%, lower in every seed, where the uncalibrated budget gives 7.6%; over-refusal stays at most 2% on 400 benign queries and MMBench is unchanged. The gain is largest where the intent sits in the image (10.1% -> 5.1%) and grows on the unseen HADES benchmark (28.9% -> 12.8%). Order-preserving distortions of the scores reproduce each predicted controller state in every seed, and the results hold under a map fitted on another judge's labels, an image-reading judge and a human audit.","authors":["Bo Han"],"cats":["cs.AI","cs.CV"],"primary":"cs.AI","published":"2026-11-12T11:00:14Z","topic":1},{"id":"2611.16627","title":"Folio: Recoverable Multi-Resolution Context Folding for Long-Horizon Language Agents","summary":"Long-horizon language agents accumulate tool results and replies until the history no longer fits the context window. Current agents shrink this history by having a model summarize it, so each compaction rereads and rewrites the history it replaces, and cache-aware triggers decide when to compact while taking the size of the result as given. We present Folio, a context manager that records, folds, and recovers history through one set of handles. While the agent runs, Folio records three linked levels: a short turn note the model writes as part of its ordinary reply (L1), an event manifest of every call written by the runtime (L2), and the captured source records (L3). Because these views already exist, the runtime folds the context programmatically, selecting coarser views for older turns. The fold prints the handles that index L2 and L3, so recovery tools that read bounded pages, compute over full text in files, and keep findings in a persistent dossier reach the evidence the agent needs with little added input. The fold trigger prices an explicit landing point that includes the retained budget. On 75 LOCA-bench tasks with DeepSeek-V4-Pro behind every system, Folio solves 66 tasks against 54 for the same runtime with model-written compaction and 49-59 for Claude Code, Codex, and SoL-Pi, in about half their working time, at 2.4 times lower cost than that runtime; each fold takes a median of 0.6 seconds.","authors":["Yubo Ruan","Tao Peng","Jiao Zhang","Chuanguo Ma","Lei Xu","Han Zhou","Wenwu Wang","Xiafei Lv","Xingjie Zhou","Guiju Fan","Hao Wang","Fenghui Zhang","Mao Miao","Shuo Li"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T10:59:40Z","topic":0},{"id":"2611.16623","title":"DySR-GNN: Dynamic Sensor-Relationship Graph Neural Network for Root Cause Analysis in Automotive HiL Multivariate Time Series","summary":"Root-cause analysis in automotive Hardware-in-the-Loop (HiL) systems is challenging because faults can propagate across multiple interacting signals, making the most abnormal sensor different from the actual fault origin. This work proposes DySR-GNN, a Dynamic Sensor-Relationship Graph Neural Network for supervised multi-label root-cause identification in multivariate HiL time series. The approach jointly extracts temporal patterns within individual sensor signals and structural relationships across sensors. Each sensor is processed by a shared temporal encoder and combined with a trainable sensor-identity representation, from which a sparse relationship graph is reconstructed independently for every input window. This window-specific structure provides an explicit representation of the sensor relationships used for graph message passing and direct root-location prediction. Under a recording-disjoint evaluation protocol, the proposed model outperforms CNN--LSTM, Static GCN, Static GAT, and context-based variants, achieving a recording-level Macro-F1 of 0.9524, MCC of 0.9302, and strict exact-match accuracy of 0.8519. Analysis of the learned graphs reveals both persistent and state-dependent relationships among sensor variables. The results demonstrate that extracting adaptive relational structure alongside temporal representations can improve multi-label root-cause identification in complex automotive cyber--physical systems.","authors":["Hamza Ouarrad","Mohammad Abboush","Andreas Rausch"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T10:58:55Z","topic":4},{"id":"2611.16621","title":"Studies on Sign Language Retrieval in Low-Resource Dictionary Settings","summary":"Most sign languages have no annotated corpus, only small dictionaries with one or a few example videos per sign. This study examines how well a pose encoder trained on a single sign language serves such languages without any target-language training. An encoder trained with metric learning on American Sign Language (ASL-Citizen) is evaluated frozen on six datasets of Turkish, Russian, Indian, Argentinian, Pakistani and Sri Lankan sign languages, with one or four clips per sign enrolled and, where signer identities are available, the query's signer excluded from the gallery. With one enrolled clip, the encoder retrieves the correct sign first in 36.4-80.7 % of queries, against 16.0-69.6 % for the best learning-free method (angle-based dynamic time warping) and a six-target mean of 39.3 % for an ASL-trained OpenHands graph network. In a crossed ablation of the encoder (embedding geometry, aggregation of body and hand embeddings, training objective, and attention between articulators), two choices matter: a per-pair InfoNCE objective outperforms supervised contrastive and cross-entropy training, and blocking attention between body, left hand and right hand inside the encoder raises accuracy by up to 5.1 points. The best configuration reaches 68.0 % one-shot accuracy (59.0 % for the reference encoder). On CISLR, an Indian Sign Language dictionary of 4,765 words with one video per word, the encoders find the correct word first for 17.3-19.0 % of queries, above the 16.8 % of the original RGB baseline. Over eight seeds with the joint encoder, spherical, Euclidean and hyperbolic embedding spaces differ by at most 1.6 points, except the hyperbolic token-attention head trained with InfoNCE, which is 1.4-2.0 points worse, so the geometry is a minor factor.","authors":["Ushnish Sarkar","Vineet Kumar Rakesh","Bhaswar Chattopadhyay","Tapas Samanta"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T10:57:50Z","topic":6},{"id":"2611.16612","title":"Unibuc - NLP at \"Voight-Kampff\" Generative AI Detection PAN 2025","summary":"This paper describes the approach of the Unibuc - NLP team in tackling the \"Voight-Kampff\" Generative AI Detection task from PAN at CLEF 2025. In the Subtask 1: Voight-Kampff AI Detection Sensitivity we have explored different approaches of leveraging embeddings extracted from LLMs, using either classical ML models or further fine-tuning LLMs with a classification head for the downstream binary classification task. Our experiments showed that using a majority vote ensemble of classical ML models (Light GBM, XGB, Logistic Regression and SVM) is comparable with fine-tuning for downstream task achieving around 99.4% F1 score. For the Subtask 2: Human-AI Collaborative Text Classification we explored a way of combining features at different layers extracted using transformer backbone with layer-wise projection and attentive pooling which proved modest results.","authors":["Teodor Marchitan","Claudiu Creanga","Liviu P. Dinu"],"cats":["cs.CL"],"primary":"cs.CL","published":"2026-11-12T10:49:03Z","topic":6},{"id":"2611.16609","title":"A Dual-Branch Graph and Convolutional Network for Interpretable Movement EEG Classification","summary":"Decoding upper-limb movements from scalp electroencephalography (EEG) is an important problem for brain-computer interfaces (BCIs), but many deep learning models provide limited insight into the brain regions underlying their predictions. We propose a two-branch architecture that combines an LGGNet-inspired local-global graph branch with an EEGNet-style convolutional branch for movement-versus-rest classification. The graph branch groups 61 EEG electrodes into anatomically informed regions and learns their spatial relationships, providing a direct representation of the regions emphasized by the model. We evaluate the proposed architecture using leave-one-subject-out cross-validation on the Ofner upper-limb movement dataset. The model achieves a mean F1-score of 0.74 +/- 0.06 across subjects. Beyond classification performance, we analyze the learned graph structure to identify the regions with the strongest connectivity. Across all held-out subjects, the highest connectivity is consistently observed in the Parietal Left Low region, corresponding to electrodes P1 and P3. This region is opposite to the performed right-arm movements and lies within the posterior parietal area previously associated with upper-limb movement processing. These results show that incorporating a region-based graph into the decoder can provide an anatomically interpretable representation while maintaining effective movement classification, offering a practical approach toward more transparent EEG-based BCIs.","authors":["Maryam Ostadsharif Memar","Behzad Nazari"],"cats":["cs.AI","cs.HC","cs.LG","eess.SP"],"primary":"cs.AI","published":"2026-11-12T10:47:37Z","topic":1},{"id":"2611.16608","title":"Towards Democratising Low-Data Medical Image Classification: CPU-Trainable ZACH-ViT and Its Geometric Variants","summary":"Medical image classification in resource-constrained settings must contend with limited annotations and computational capacity. We evaluate four ZACH-ViT variants with linear, Euclidean prototype, Poincare and spherical heads alongside two compact multiple-instance-learning baselines. All models are trained from scratch with fewer than 260,000 parameters across seven MedMNIST+ datasets and three external benchmarks (PCam, ISIC2017 and Shenzhen tuberculosis radiographs). The low-data protocol uses 50 training images per class and five seeds, with validation-selected checkpoints at 23 and 47 epochs; a full-data reference uses one seed. Compact-TransMIL achieves the highest ten-dataset mean test Macro-F1 at 23 epochs (53.34%) and with full data (73.93%). At 47 epochs, ZACH-Linear narrowly leads (57.76% versus 57.54% for Compact-TransMIL). Extending training improves Macro-F1 in 239 of 300 paired comparisons. On ISIC2017, hZACH-Spherical has the highest 47-epoch low-data Macro-F1 (44.09% with SD 4.09%), whereas ZACH-EuclideanProto leads with full data (51.27%). We additionally conduct 40 CPU-only training runs on ISIC2017 and Shenzhen using four CPU threads; model-only single-image inference requires approximately 4-9 ms, while image decoding and preprocessing substantially increase end-to-end latency. The results demonstrate the feasibility of CPU-only training on the tested platform and the task- and regime-dependence of compact medical imaging under strict resource constraints.","authors":["Athanasios Angelakis"],"cats":["cs.CV","cs.LG"],"primary":"cs.CV","published":"2026-11-12T10:46:53Z","topic":3},{"id":"2611.16591","title":"Singularity-Bench: Can Agents Build Agents?","summary":"Frontier AI agents increasingly succeed on realistic, long-horizon tasks and are beginning to improve the systems through which they operate. Yet existing benchmarks often restrict agents to narrow action spaces, use simplified environments, or rely on verification methods that are unreliable for evaluating arbitrary agent-generated systems. We introduce Singularity-Bench, an open-ended benchmark for autonomous AI development with six end-to-end tasks spanning pre-training, mid-training, post-training, inference, the agent harness, and cluster orchestration. Each task requires agents to implement core systems from a minimal starting point and evaluates them against mature human-engineered baselines on identical hardware. Additionally, through extensive red-teaming covering 9 of the 10 categories in the OWASP Top 10 for Agentic Applications, we expose reward-hacking vulnerabilities in the conventional Harbor framework and develop Breakwater, an extension of Harbor that blocks all attacks tested in our study. Across 11 agents built from 7 frontier models and 2 harnesses, Opus 5.5 with Claude Code matches or outperforms the human baseline on all six tasks with a Singularity-Index of 1.25, suggesting it may have the skills needed to reach the singularity. The remaining agents average only 0.58. Our vision is for Singularity-Bench to track how much of AI development agents can already automate and when they may transition from assisting to taking end-to-end responsibility for building stronger successors.","authors":["Yang Zhou","Krish Agarwal","Zhuoming Chen","Zerui Wang","Zhenyu Zhang","Wenshuang Li","Xinpeng Wei","Zhichen Zeng","Jingyu Liu","Jue Wang","Sen Wu","Ziheng Jiang","Zhen Dong","Han Cai","Binhang Yuan","Ying Sheng","Ce Zhang","Bo Li","Beidi Chen"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T10:24:51Z","topic":0},{"id":"2611.16590","title":"AstroMemBench: A Benchmark for Memory Systems in Embodied Spacecraft Agents","summary":"Long-term spacecraft autonomy requires embodied agents to use operational history under communication delays, time-critical operations, environmental uncertainty, complex physical interactions, power and computational constraints, and safety and reliability requirements. We introduce AstroMemBench and a Scenario-Challenge-Capability design method that maps these requirements to executable memory tests. The benchmark covers 16 memory subchallenges through 140 templates and 2240 instances across nine tasks in planetary surface operations, on-orbit servicing, and deep-space exploration. A deterministic environment scores actions for task completion, state changes, safety, and resource use. Ablations under nine conditions show that the benchmark is sensitive to how accurately agents recover and use historical information and to differences in memory-system design. Main experiments compare six configurations. Mem0, Chain-of-Memory, LightMem, and retrieval-augmented generation (RAG) improve same-session mean scores over native Pi Agent, with no clear advantage of the three memory systems over RAG. After a session restart, Mem0 outperforms RAG by 12.25 points. Benefits favor instruction retention and result reuse over ongoing state tracking and energy reservation. AstroMemBench supports the design, development, and selection of memory systems for embodied spacecraft agents. AstroMemBench 1.0 and its evaluation code are publicly available: https://github.com/flyingView/AstroMemBench-v1.","authors":["You Wu","Guhong Zhang","Hongwei Li","Xinyue Hong","Yuan Jia"],"cats":["cs.AI","cs.RO"],"primary":"cs.AI","published":"2026-11-12T10:23:15Z","topic":0},{"id":"2611.16586","title":"Large Language Models as Digital Twins: a Distribution-Aware Questionnaire-Based Evaluation","summary":"This paper investigates whether Large Language Models can answer a personality questionnaire as a specific individual would. Given a user's social-media posts, the model completes the BFI-2-60 Big Five questionnaire as that user. The predicted responses are converted into trait scores and compared with the user's actual scores. We evaluate GPT-4, Gemma, Llama 3.1, and Mistral on the MyPersonality dataset. To determine whether predictions reflect information from the user's writing, we compare them with answers produced without user text and with values derived from general population tendencies. We measure prediction error across all users and separately for individuals with low, intermediate, and high trait scores. We also test whether an LLM can express an explicit personality profile when given only the target Big Five scores, without user text, and examine how predictions change as more text is provided. The results show that user text does not consistently improve performance. Models perform better near the middle of the trait distribution, while at the lower and upper ends they often remain close to central values or perform well at one end but poorly at the other. When given an explicit profile, GPT-4 most clearly translates trait information into consistent questionnaire responses, particularly for Extraversion and Neuroticism. Increasing context does not systematically reduce error: open-source models fluctuate more, whereas GPT-4 remains comparatively stable. Overall, the evaluated models do not consistently reproduce the questionnaire behavior of specific individuals.","authors":["Francesco Di Cursi","Chiara Boldrini","Andrea Passarella","Marco Conti"],"cats":["cs.AI","cs.CY"],"primary":"cs.AI","published":"2026-11-12T10:18:16Z","topic":6},{"id":"2611.16581","title":"Do the Details Help the Whole? One-Way Transfer from Person Actions to Group Intent in VLMs on JRDB-Social","summary":"Social scene datasets label people at two levels: what each person is doing and what the group is there for. Training on both levels at once is often assumed to help each of them. We test this assumption for vision-language models (VLMs) on JRDB-Social, treating person actions as the details and group intent as the whole. We fine-tune Qwen3.5 4B under eight training arms with three seeds, repeat the core arms on Qwen3.5 9B and Qwen3-VL 4B, and judge every arm with a verdict rule fixed before testing. Writing each member's actions before the intent raises intent macro-F1 from 30.1 to 31.0, and writing the intent first reaches 31.2, while action F1 stays within 0.2 points of an action-only model. Under the selected recipe, no arm trained with intent improves actions. The gain requires correct, image-grounded actions: empty action lists or actions borrowed from another group with the same intent lower intent F1. Forced-prefix tests, occlusion, member masking, linear probes and component grafting show that the intent follows the written actions, relies more on people and partly resides in the vision tower. Supplying actions as input instead of as targets hurts both levels.","authors":["Chengguang Gan","Zhixi Cai","Yunhao Liang","Hanjun Wei","QingHao Zhang","Shiwen Ni"],"cats":["cs.CV","cs.CL"],"primary":"cs.CV","published":"2026-11-12T10:14:07Z","topic":3},{"id":"2611.16574","title":"HALight: Heterogeneous Multi-Agent Reinforcement Learning for Traffic Signal Control under Lane-Level Missing Observations","summary":"Multi-agent reinforcement learning has shown great promise for traffic signal control. However, most existing methods rely on two key assumptions: (i) agents are assumed to be homogeneous with unified observation/action spaces; (ii) full and reliable lane-level traffic observations are available to all agents. In practice, urban road networks exhibit substantial structural heterogeneity (e.g., lane configurations and phase designs), while sensor failures and incomplete detector coverage often result in missing incoming lane measurements. Such issues can severely impair coordination and destabilize learning. To address these challenges, we propose HALight, a communication-enhanced heterogeneous multi-agent reinforcement learning framework for traffic signal control under lane-level missing observations. HALight supports intersection-specific observation and action modeling to accommodate heterogeneous intersections, and incorporates a lightweight one-hop neighborhood communication module to enrich spatial context when lane-level observations are partially unavailable. Moreover, HALight adopts a sequential policy update scheme to improve coordination among heterogeneous agents. Experimental results on two real-world road networks demonstrate that HALight outperforms representative homogeneous and heterogeneous baselines in terms of traffic efficiency and convergence behavior. The code and datasets used in this work are available at https://github.com/Huchenkaii/HALight.","authors":["Chenkai Hu","Yingqian Zhang","Ming Xie","Yi Fang","Chengxiang Jin","Chao Li"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T10:02:01Z","topic":0},{"id":"2611.16565","title":"Look Here, Then Decide: Small Decision Models in a Harness Find Memory-Safety Vulnerabilities at Scale on Commodity GPUs","summary":"We ask whether small, open-weight decision models can support vulnerability discovery in open-source software at scale and what the surrounding system must provide. These models answer a typed question in one forward pass, returning scores for the permitted answers rather than text. Our harness selects widely packaged C and C++ projects and uses deterministic incomplete-fix miners and shape localizers to decide where to look. Intern-Decision-4B, a 4B-parameter open-weight decision model, then scores every candidate function. The scores rank the queues rather than accept or reject candidates, apart from a low reach-score floor that some queues kept. The search uses no frontier model. Alone, the model is weak: it separates vulnerable from benign functions with AUC 0.73, but a vulnerable function from its own patch with only 0.56. It is more useful for ordering candidates. On 75 public CVEs from projects without any of our findings, the fixed function reached its repository's top 25 in 21% of cases, 1.4 times as often as under random order of the same queues. Recall is bounded mainly by where the localizers look. On the campaign's own queues the model's order placed 7.5 times more confirmed findings in the top 10 than random order, an observed, selection-biased upper bound. On three GB10 systems, the harness searched more than 4,000 projects at 900-1,100 repositories per hour. The campaign confirmed 187 memory-safety findings; by worst-case CVSS v3.1, 12 are critical, 148 high, 23 medium and 4 low. The decision step runs at 4.4 decisions per second on one GB10 and 9.8 on a consumer RTX 4090 gaming PC, with matching scores. A small decision model supports vulnerability discovery through prioritization, while deterministic analysis and verification remain essential. All confirmed findings are being disclosed to maintainers under coordinated disclosure before publication.","authors":["Tim De Wachter","Kyryl Volkov","Myrto Eleftheria Gkogkou","Mathieu Huysman"],"cats":["cs.CR","cs.LG","cs.SE"],"primary":"cs.CR","published":"2026-11-12T09:51:23Z","topic":1},{"id":"2611.16561","title":"SPARC: Accurate Photonic Matrix Multiplication via Representative Vectors","summary":"Photonic neural network accelerators have emerged as a promising solution for reducing the power consumption of deep neural network (DNN) inference by executing matrix multiplication on analog photonic hardware. However, their inference accuracy is limited by low-bit quantization required for photonic processing and by analog photonic noise. This paper proposes an approach for improving inference accuracy by exploiting the high similarity among activation vectors in DNNs. Specifically, our approach constructs a representative vector for each group of similar activation vectors, computes its inner product digitally with high precision, and offloads the residual inner products to the photonic accelerator. Since the activation vectors within each group are highly similar, the residual vectors have much smaller norms, thereby reducing the effects of quantization errors and analog photonic noise. Furthermore, our approach preserves the photonic inner-product computation pipeline, allowing seamless integration into existing and future photonic accelerators without hardware modification. Experimental results on six representative DNN models demonstrate that the proposed approach can improve inference accuracy across diverse network architectures and application domains.","authors":["Yasuhiro Fujiwara","Tatsuhiro Nakamori","Hideyuki Kawashima"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T09:47:42Z","topic":4},{"id":"2611.16560","title":"When and How SignSGD Outperforms SGD: A Functional Scaling Law Perspective","summary":"We study how the sign operator interacts with gradient noise to shape the learning dynamics and scaling behavior of sign-based optimizers. In power-law kernel regression with diagonal features, we establish a functional scaling law (FSL) for SignSGD under general learning rate schedules. The FSL describes the learning dynamics through two processes: signal learning and noise forgetting. The key mechanism is noise-induced preconditioning: gradient noise smooths the sign operator, inducing an effective preconditioned SGD. Noise-curvature alignment enables this preconditioning to transform the effective Hessian from $\\mathbf{H}$ to $\\mathbf{H}^{1/2}$. The resulting dynamics exhibit faster signal learning but slower noise forgetting than SGD. This tradeoff yields a regime-dependent comparison with SGD: (i) SignSGD achieves improved data efficiency in the hard-task regime; (ii) both optimizers achieve the same data efficiency in the easy-task regime, but SignSGD maintains this efficiency at larger batch sizes, requiring fewer optimization steps for a fixed data budget. Power-law kernel regression experiments support the proposed mechanism and predicted scaling behavior. LLM pre-training experiments further suggest that the SignSGD FSL better captures the batch-size dependence of loss dynamics for sign-based optimizers than the original SGD FSL.","authors":["Zilin Wang","Binghui Li","Jia-Nan Wang","Lean Wang","Jinbo Wang","Lei Wu"],"cats":["stat.ML","cs.LG"],"primary":"stat.ML","published":"2026-11-12T09:47:04Z","topic":1},{"id":"2611.16555","title":"Similarity-Guided Optimization Strategies for Approximate Network Symmetry Detection","summary":"Symmetry in complex networks is a structural property with broad implications for network function, dynamics, and robustness. While classical approaches rely on exact automorphisms and thus remain highly sensitive to small perturbations, recent work of Liu has introduced approximate formulations that relax the strict invariance condition and recast symmetry detection as an optimization problem over permutations. We investigate how node-level similarity information can be used to bias the search toward structurally plausible vertex correspondences, with network centralities serving as one interpretable instance. Building on this perspective, we adapt simulated annealing, genetic algorithms, tabu search, particle swarm optimization, differential evolution and the quadratic symmetry approximator to this problem, add similarity guidance to each, and introduce Bilinear Approximate Symmetry Inference (BASIN), which learns a soft assignment matrix via bilinear embeddings. Using a unified evaluation pipeline, we compare their guided and unguided variants, BASIN and a random baseline on four synthetic graph families, structural brain connectomes, and three empirical networks with known automorphism groups. Node-similarity guidance is decisive where a planted orbit exists that the unguided search misses, and algorithm-dependent elsewhere, on synthetic and empirical networks alike. Overall, tabu search achieves the strongest performance; its guided variant (SGTS) has the lowest median score among the guided variants in all 44 synthetic conditions, while BASIN is the better of the continuous relaxations on rewired lateral graphs. The effect of node-similarity guidance is therefore regime-dependent rather than uniformly positive.","authors":["David Hartman","Jaroslav Hlinka","Dominik Klepl","Anna Pidnebesna","Petra Vidnerová"],"cats":["cs.NE","cs.LG","q-bio.NC"],"primary":"cs.NE","published":"2026-11-12T09:44:02Z","topic":1},{"id":"2611.16527","title":"AI has its own style in mathematical proving","summary":"As large language models increasingly generate mathematical proofs, whether they develop a proving style distinct from that of human mathematicians remains an open question. This question is particularly pressing as mathematicians increasingly voice concerns that proofs generated by large language models are difficult to interpret. Here we ask three large language models to independently reprove all 465 propositions of Euclid's \\textit{Elements} and compare their citation choices against the human-proof baseline through Jaccard similarity and complex network analysis. We find that AI-generated proofs differ from human proofs, and this difference is stable and reproducible: mean pairwise Jaccard similarity among AI runs (~0.68) is nearly twice that between AI and human proofs (~0.37), and cross-run classification consistency is also high. By examining citation structure in detail, we identify the concrete signature of this difference: approximately 40% of AI citations point to foundational elements (definitions, common notions, and postulates), compared with only 13\\% in human proofs; even among theorem-to-theorem citations, AI tends to cite more distant results. Roughly half of AI's extra citations trace to the upstream dependencies of the human proof. Together, these findings suggest that the AI proving style is a tendency to construct arguments on its own from foundational elements rather than directly citing later, more established results, which we term \"building from the ground up\". Our findings offer a new perspective on how AI systems organise and invoke mathematical knowledge, with implications for understanding the interpretability of AI-generated proofs.","authors":["Xiaokai Li","Ziyu Zhou","Qibin Cheng","Yangcheng Wu","An Zeng","Ying Fan"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T09:27:02Z","topic":5},{"id":"2611.16526","title":"Lost in Delegation: Studying the Effectiveness of Task Orchestration in Generative Search","summary":"LLMs have quickly evolved from models used in isolation to agentic systems that delegate tasks to external tools and other LLMs to solve complex assignments. This agentic augmentation has led to impressive capability gains. However, when and how delegation improves performance remains poorly understood. For example, it is unclear whether delegation helps when tasks involve post-cutoff information that the orchestrating LLM cannot reliably resolve from its own knowledge. In this work, we use purpose-built datasets with pre- and post-cutoff splits to compare various delegation modes. We find that delegation provides no consistent accuracy improvement over non-delegated setups, with effects varying across models and tasks. When tasks involve knowledge unavailable to the orchestrator, delegation can substantially degrade accuracy, while direct tool use with the same LLM performs much better. Our analysis identifies distorted delegated queries, inefficient tool use, and failures to preserve and use information between agents as recurring failure modes. These findings suggest that effective orchestration requires both well-calibrated delegation decisions and effective exchange of information between agents.","authors":["Tamara Stojanovska","Elisabeth Kirsten","Leon Swazinna","Muhammad Bilal Zafar"],"cats":["cs.AI","cs.CL"],"primary":"cs.AI","published":"2026-11-12T09:25:33Z","topic":0},{"id":"2611.16525","title":"ZONOS3-Mini Technical Report","summary":"ZONOS3-Mini is an ultra-compact, ultra-fast, zero-shot voice-cloning TTS model. The model can run in real time on a laptop or smartphone. For its latency and size, it achieves state-of-the-art accuracy and cloning fidelity. To achieve this we propose a latent MoE backbone with only 886M parameters (100M active per token) and a new audio codec, Zyphra Audio Codec 1 (ZAC-1), which represents 44.1 kHz audio at a 15 Hz frame rate. On the new ZTTS2-Eval benchmark, ZONOS3-Mini reaches a 6.71% macro-average WER over 55 language sets (35 clean and 20 in the wild), which is competitive with larger and slower state-of-the-art models. We release the model weights, inference, and training code.","authors":["Gabriel Clark","Sofian Mejjoute","George Close","Beren Millidge"],"cats":["cs.SD","cs.AI"],"primary":"cs.SD","published":"2026-11-12T09:24:22Z","topic":3},{"id":"2611.16524","title":"A Joint LLM and Optimization Agent Solution for Drone Scheduling in Victim Search Missions","summary":"Victim search in wilderness areas is time-critical. Drones enable rapid area search, but scheduling them is challenging because victims may continue moving, detection results are uncertain, and effective decisions must combine structured mission information with unstructured contextual clues and domain knowledge. To address this, we propose an agent-aided drone scheduling system. The system employs an agent at the ground control station (GCS) to coordinate multiple drones over successive search cycles. In each cycle, the agent applies a scheduling policy that combines a large language model (LLM) for reasoning about where to search with an optimizer for determining how to dispatch the drones under energy constraints. Simulation results show that the proposed system consistently completes the victim-search mission, whereas none of the six non-LLM baselines does so and direct LLM scheduling is considerably less reliable. The source code is publicly available at https://github.com/1643204431/llm-uav-victim-search.","authors":["Yinghao Zhang","Yalin Liu","Hong-Ning Dai","Yan Zhang","Qubeijian Wang","Bishenghui Tao","Yaohan Li"],"cats":["eess.SY","cs.AI","cs.ET"],"primary":"eess.SY","published":"2026-11-12T09:20:54Z","topic":0},{"id":"2611.16523","title":"RESYNC: Residual Latent-State Synchronization for Deadline-Bound Streaming Inference","summary":"Predicting every observation in a high-rate stream requires a model to finish before the next arrival. A large model may offer better predictions but miss that deadline. We propose RESYNC: the large model runs asynchronously, while a fast path updates its latest exact but stale latent state using the observations received since that state was computed and applies its unchanged prediction head. The updater learns from the frozen model's representations without task labels. Across one real and two synthetic streams and five architecture families, RESYNC beats the best compact or distilled alternative with a paired 95% interval above zero in all twelve dataset--architecture pairs with a material teacher advantage at anchor ages of 4--12 observations, and in eleven pairs at older tested ages up to 32. The share of the teacher's gain preserved remains complete on the real stream but is lower and generally declines with anchor age on the synthetic streams. In live replay with 1 ms arrivals, the synchronous large models cannot deliver any predictions on time while RESYNC serves at least 99.4% of arrivals within the interval, with 0.35--0.62 ms median latency on one CPU thread: around the same speed as or faster than the compact model, and 7.9--17.7$\\times$ faster than synchronous large-model inference. RESYNC targets settings with background compute available but a strict per-observation response deadline.","authors":["Eljas Linna","Hao Dong","Juho Kanniainen"],"cats":["cs.LG","cs.AI"],"primary":"cs.LG","published":"2026-11-12T09:19:39Z","topic":4},{"id":"2611.16522","title":"PISim: A Unified Photonic Inference Simulator for Hardware-Aware Cross-Layer Design-Space Exploration","summary":"Photonic neural accelerators offer ultra-low latency linear algebra, yet their usable numerical precision, sustainable throughput, and application accuracy remain tightly bounded by surrounding electro-optical interfaces and physical non-idealities. Pre-fabrication design-space exploration (DSE) requires a cross-layer electronic design automation (EDA) framework capable of evaluating disparate optical topologies under identical neural workloads and physical assumptions. We present \\textbf{PISim}, a unified Photonic Inference Simulator that couples pretrained deep neural network execution with physics-based models of six accelerator architecture families, six electro-optic modulator technologies, four optical source organizations, and direct or coherent readout schemes. PISim embeds stage-resolved impairment operators---covering control and analog-to-digital quantization, finite extinction ratio, shot noise, correlated relative-intensity noise (RIN), thermal drift, propagation loss, and channel crosstalk---alongside an analytical cycle-level interface timing model to expose electronic-optical bottlenecks. Using simulated MNIST multilayer perceptron (MLP) benchmarks, we demonstrate how PISim identifies optimal precision operating points under multi-impairment stress (peaking at 6 bits for Mach--Zehnder and microring modulators), quantifies a 52.5 percentage-point accuracy gap between VCSEL-array and parallel-fanout laser architectures arising, under the modeled conditions, from power partitioning and correlated source noise, and models interface-limited throughput scaling across emitter-array widths from 32 to 128 channels. This comprehensive framework establishes an extensible, reproducible foundation for accuracy-constrained Pareto optimization prior to photonic tape-out.","authors":["Andreas Falkenberg"],"cats":["physics.optics","cs.AI"],"primary":"physics.optics","published":"2026-11-12T09:18:27Z","topic":4},{"id":"2611.16514","title":"Learning Temporal Dynamics with Derivative Representations for Time Series Forecasting","summary":"Time series forecasting aims to predict the future evolution of complex systems by learning their underlying dynamics from past observations. Despite recent advances in deep learning, Transformers in particular still struggle to capture the temporal evolution of system dynamics. Existing approaches have been largely driven by architectural innovation, with comparatively little attention paid to training strategies that explicitly exploit the temporal structure governing system dynamics. In this work, we introduce DeltaTSF, a dynamics-aware forecasting framework based on discrete derivative representations of time series. Specifically, DeltaTSF jointly learns the series and their discrete temporal derivatives across both input and forecast horizons within a unified, model-agnostic framework. By coupling these representations over time, the model is encouraged to maintain temporal consistency, a property we show to induce a form of discrete Taylor consistency during training. This formulation enables the model to better reflect the underlying system dynamics, resulting in more stable and physically coherent forecasts. Notably, DeltaTSF does not require explicit derivative computation and introduces no additional computational overhead at inference time. Extensive experiments across diverse benchmarks and backbone architectures demonstrate that DeltaTSF consistently improves forecasting performance and yields more dynamically coherent predictions than strong SOTA methods. These results highlight the effectiveness of enforcing dynamic coherence through derivative representations as a model-agnostic paradigm that operates alongside and complements architectural design. Source code will be made publicly available upon acceptance.","authors":["Shujie Duan","Jia Hong Puah","Haocheng Wu","Yifei Mao","Hongting Zhou","Chenxuan Li","Zitao Fang","Yang Lu","Sim Kuan Goh"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T09:12:03Z","topic":6},{"id":"2611.16502","title":"SignRot: LLM Quantization with Massive Outlier-Aware Sign-Adjusted Rotation","summary":"While quantization compresses LLMs and accelerates inference, outlier activations impede efficient low-bit representation. In particular, massive outliers cause substantial performance degradation. Although rotation-based methods that rely on randomized Hadamard transforms for online inference aim to mitigate this, they can produce fragmented distributions with empty bins when a massive outlier dominates a token's activation. This fragmentation severely undermines uniform quantization performance. Moreover, massive outliers concentrate in the down-projection input, where the preceding nonlinearity requires the rotation to be applied at inference time. This constraint prevents training-based approaches such as SpinQuant from learning the online rotation at this position, as a learned dense matrix cannot use the Fast Walsh-Hadamard Transform and would require a full matrix multiplication during inference. To address this, we propose SignRot (Sign-adjusted Rotation), which identifies massive outlier channels and constructs a sign adjustment vector that unifies the signs of the corresponding Hadamard row, collapsing the bimodal distribution into a unimodal profile amenable to quantization. Since the aligned outlier shifts the activation mean away from zero, we pair SignRot with asymmetric activation quantization, whose zero-point absorbs this shift. As a lightweight, drop-in improvement compatible with any Hadamard-based quantization method, SignRot requires only a single calibration forward pass and adds only element-wise sign flips at inference, without an additional matrix multiplication. Applying SignRot to QuaRot achieves a WikiText2 perplexity of 5.88 and an average zero-shot reasoning accuracy of 72.73, surpassing SpinQuant on the LLaMA3-70B model. Code is available at https://github.com/HYU-ARK-Lab/SignRot.","authors":["Jaehun Gim","Jaewoo Kim","Gyuwan Kim","Seo Yeon Park"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T09:05:14Z","topic":4},{"id":"2611.16500","title":"EvenVLM: A Text and Multimodal Dataset for the Even Language with Vision-Language Fine-Tuning","summary":"We present EvenVLM, the first text and multimodal (VQA) dataset for Even (Lamut), an endangered Northern Tungusic language spoken by approximately 6,800 people in Siberia. Our dataset comprises 50,521 structured Even-Russian entries extracted from dictionary and phrasebook sources via GPT-4o Mini-based parsing, from which we synthetically generate 107,543 training examples spanning translation, grammar, reading comprehension, and semantic classification. To address tokenization inefficiency on Even's agglutinative morphology, we propose a three-step tokenizer expansion algorithm that adds Even-specific graphemes (eng, oe, o-umlaut), whitelisted vocabulary, and BPE merges, reducing average tokens per word by 27% (4.99 to 3.64). Using LoRA fine-tuning of Qwen3-VL-4B, we adapt the model for text tasks (0.53% trainable parameters, rank 32) and separately for visual question answering (rank 64, frozen vision encoder). Our best configuration achieves 11.4% exact match and a 13.6 weighted score on text benchmarks, 0.1283 chrF for sentence translation, and 65% accuracy on VQA (up from 0% for the base model and text-only fine-tuning). We release our datasets, evaluation benchmarks, and training pipeline to support further research on Even and other low-resource languages.","authors":["Daniil Makovetskii","Ivan Bondarenko"],"cats":["cs.CL"],"primary":"cs.CL","published":"2026-11-12T09:03:06Z","topic":3},{"id":"2611.16499","title":"Recalibrating a Typed-Decision Model in an Afternoon: A Hands-On Recipe for Laya's Confidence Scores","summary":"A typed-decision model such as Laya reads a text and a list of options that the caller defines, for example ticket categories, and returns a probability for each option. Code around it compares the top probability with a threshold to answer, abstain or escalate to a person. Earlier work reports, for a Turkish model, that temperature scaling can raise held-out calibration error (ECE). This tutorial gives a five-step recipe (measure, diagnose, fit, validate, deploy) in 130 lines of numpy with standard methods, none new. On 1,065 synthetic, LLM-labelled items the released Laya checkpoint has a pooled ECE of 0.073 but a bucket-macro ECE of 0.143 (0.127-0.184), the mean over buckets (questions grouped by type and option count): the 12-option bucket is over-confident by +0.22, the three-to-five-option bucket under-confident by 0.05, and they cancel. One global temperature makes pooled ECE worse (0.173); a cross-fitted temperature per bucket brings macro-ECE to 0.073 (0.078 over 10 fold assignments) and reaches 0.089 with 10 labelled items per bucket, but holding out one of seven evaluation sets gives 0.099. After the refit, a gate aimed at 10% error answers 46.3% of items (10.5% wrong) against 6.9% on the published confidences, but one cut-off per bucket on the published confidences answers 45.4%, so most of the gain comes from separating buckets. Confidence-only maps (Platt, histogram, isotonic) need 20 or more labelled items per bucket to approach the temperature; with 320 per bucket the lowest macro-ECE belongs to Platt. In these files the retrained checkpoints store no per-bucket table and the stored 12-option value 0.1006 is clamped to 0.5 at runtime; the refit helps 31 of 33 checkpoints on random folds and 21 with a set held out. Fit and test share one generator and LLM labels, so these are findings about these checkpoints and this data.","authors":["Mohammed Nayeem Khan","Aliyas Ahmad Misri"],"cats":["cs.AI","cs.SE"],"primary":"cs.AI","published":"2026-11-12T09:02:57Z","topic":6},{"id":"2611.16498","title":"How many LLM-labelled rows does a decision encoder need, and how wrong can the labels be?","summary":"A decision encoder such as Laya reads a question with named options (for example, which department should handle a ticket) and a text, and picks an option. It reads both in one forward pass and returns a probability for every option, without writing any text. We ask how many LLM-written training rows such an encoder needs and how wrong those labels may be. We ran 19 fine-tunes of Laya (ModernBERT-large) on nested subsets of a 2,818-row pool and scored them on 1,065 held-out items from the same generator, so every test item is in-distribution and no out-of-family test was run. Accuracy rises from 69.4% before fine-tuning to 74.3% with 5% of the rows, 84.8% with 50% and 86.9% with all rows. The curve flattens, and a saturating fit with bootstrap intervals cannot pin down its plateau. Seed spread at 20% of the rows (s.d. 1.6 points) exceeds some of the steps between sizes. Giving small subsets as many updates as the full pool gains 1.9 points at 5% and 1.9 points at 20% of the rows (one seed; p=0.14 and 0.13), and at 20% that is below the 2.3-point difference between two ordinary seeds. A single run with 10% symmetric label noise shows no detectable change (-0.2 points, 95% interval -1.7 to +1.4); 20% costs 3.4 points and 40% costs 8.7 points. Confusable-pair noise at 20% costs 7.4 points, and under noise the model moves from over-confident to under-confident. An independent labeller disagrees with the gold labels on 3.5% of held-out items, and about half of its choice disagreements (16 of 33, leaving each item out of the partner map) name the partner option that the confusable-pair noise uses. Laya is +0.9 points ahead of a TF-IDF classifier trained on the same rows at full size (the interval includes zero) and +21.1 points ahead at 5% of the rows. Everything is synthetic, LLM-labelled and from one backbone, one trainer and mostly one seed.","authors":["Mohammed Nayeem Khan","Aliyas Ahmad Misri"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T09:02:55Z","topic":6},{"id":"2611.16493","title":"ReA-Bench: Benchmarking Responsible Agent Auditing","summary":"As agents perform tasks autonomously in local and cloud environments, auditing their execution trajectories can help service providers identify risks and improve agent harnesses and foundation models. Our analysis of production and red-team trajectories shows that some risks are repeatedly missed across auditing configurations. Overall recall obscures which risks are missed and whether higher recall comes with more false positives on legitimate behavior. Existing benchmarks offer limited insight into these failures. Generated trajectories can contain inconsistent tool calls and responses, conspicuous risky actions, or repetitive attack patterns. Grouping results by risk type also makes it difficult to tell whether auditors overlooked evidence, accepted an unsupported conclusion, or lacked information needed to make a judgment. We introduce ReA-Bench, a diagnostic benchmark for auditing responsible agent behavior. It organizes auditing challenges into six tiers: internal safety knowledge, contradictions within trajectories, evidence retrieval in long contexts, verification of sources and reasoning, indirect steering, and missing information. It includes challenging benign cases in which actions resemble risky ones but are consistent with the task and its authorization. It also provides paired trajectory versions to test whether auditors detect the same risks more reliably when the trajectory is shortened or missing information is supplied. Most samples are drawn from real production trajectories, with scenarios manually selected and adapted to preserve realistic tasks and risks that could plausibly arise during execution.","authors":["Zipeng Ye","Shenyu Dai","Yong Xie","Kun Wang","Jiyang Guan","Zhikai Chen","Jiaqi Weng","Jialing Tao","Hui Xue"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T09:01:24Z","topic":0},{"id":"2611.16492","title":"Time and Space: A Dual-Axis, Evidence-Grounded Context Substrate for Personal Agents","summary":"We argue that a binding constraint for personal agents is not model capacity but the form in which context is represented: the agent does not know which entity a phrase denotes, when a fact held, or which source record licenses a claim. We call the missing representation spatio-temporal context (STC) and define it on two axes. The temporal axis carries event time, validity intervals and ingestion time. The spatial axis is a relational space in which entities are points, typed relations are edges, and distance is semantic rather than geographic; physical places are one entity type among many. Every relation is anchored to immutable source records, and every answer must be closed over those records. We formalise the substrate and six invariants, a context operator that composes temporal intersection with relational expansion, an ambiguity-preserving resolution state machine, a conflict detector that converts mutually exclusive facts into user adjudication rather than silent overwrite, and a four-regime consumption taxonomy spanning query, insight, triage and proactive reminder. We describe a local-first prototype of the retrieval layer built on SQLite with on-device embeddings and hybrid lexical-dense recall, report measurements from a deployed store of 341 documents, 340,079 chunks and 190,271 embedded vectors, and specify, but do not yet implement, the analytic layer on which the framework depends. We close with an evaluation protocol with per-invariant metrics for evidence coverage, temporal correctness, conflict precision and interruption cost. The proposal targets the representation of context rather than the volume of it.","authors":["Haitao Cheng","Jerry Hu","Longdong Mao","Dong Huang","Dongming Li","Lijuan Zhang","Mengting Zhang"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T09:01:06Z","topic":0},{"id":"2611.16487","title":"HeMeR: Heterogeneous Memory Reconciliation for Embodied Agents via Structured KV Reuse","summary":"Embodied agents make decisions by combining information from heterogeneous memory modules, including declarative memory for explicit knowledge, procedural memory for learned routines, and working memory for current observations. During interaction, the memory contents required for decision making change with the agent's state, goal, and task progress, while many contents remain shared across consecutive decisions. However, existing memory modules are typically accessed through separate interfaces, so each updated memory context is often processed as a new input to the foundation model. This prevents selective key-value (KV) cache reuse for unchanged memory contents and forces full KV cache recomputation even when much of the memory context remains valid. We introduce HeMeR, an inference framework that performs KV reconciliation between dynamically updated memory contexts and reusable KV states. HeMeR tracks declarative, procedural, and working memory contents across action decisions, preserves KV states for unchanged contents, and reconciles only the KV states affected by memory-content updates. This enables efficient and stable inference as memory contents evolve during interaction. We evaluate HeMeR on ALFWorld, Habitat, and real-world robotic deployments requiring coordinated use of heterogeneous memory contents. In the real-world setting, HeMeR achieves 83.20% task success and reduces per-step inference time from 2.94 to 1.87 seconds compared with MemRL, while preserving compatible KV states as task progress updates the memory context.","authors":["Saehun Chun","Wonje Choi","Jinwoo Jang","Taeyoon Kwack","Honguk Woo"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T08:58:20Z","topic":0},{"id":"2611.16485","title":"Contain, Conform, Compare (Extended Version): A Red-Teamed Evaluation Kit for Small Coding Agents and the Full Record of Its Experiments","summary":"A hosted score in the Gemma 4 Developer Agent competition is one noisy number, so development needs a local evaluator that is fast, safe for agent-written code, and matching the hosted scorer's prompts. drydock-eval evaluates a small coding agent (Gemma 4 E4B) at about 250 runs per hour on one RTX 5090. Contain runs each job in a fresh container without network and passed 200 scripted escape tests; conform provides eight checks that compare a local stack's requests with the scorer's configuration, down to rendered tokens; compare provides paired statistics that respect repository clusters. AI agents built and attacked the kit; a human owner approved machine use and release. This extended version repeats the short paper and adds the full record of the experiments made with and around the kit: 9,660 rows of agent runs, 2,494 grading controls and 72 serving measurements, all released. On four L4 GPUs the 31B model decodes 38.1 tokens per second in one stream and E4B 105.8. Six configurations of a 31B agent solved between 37 and 44 of 70 development tasks, and no pairwise difference was detected. A pre-registered five-arm experiment (231 tasks, 4,620 jobs) measured four prompt divergences: no E4B solve-rate change was detected (largest estimate -2.1 points, 95% interval -4.4 to +0.4). Three changes of the E4B agent led their screens by +1.0 to +6.2 points; each was then tested once on new tasks under a frozen pass rule, and none was confirmed (gains -0.35, -0.39 and +4.10 points). The unchanged baseline, run again on the same 128 tasks, moved by 1.17 points, and 65% of these tasks were solved in none of eight runs. Prompts and agent bundles are withheld; every run-level result is public.","authors":["Nguyen Minh Thang"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T08:54:03Z","topic":0},{"id":"2611.16484","title":"TAPES: Terminal Adjoint-Propagated Error Scheduling for Financial Diffusion Samplers","summary":"Financial diffusion samplers are often reused with a small number of function evaluations (NFE) inside scenario and risk pipelines, where numerical time-grid error can distort terminal quantities such as covariances, tails, and drawdown proxies. We study offline grid construction for a frozen probability-flow ODE: the score network, conditioning, Euler solver, report normalization, and deployment NFE remain fixed. TAPES (Terminal Adjoint-Propagated Error Scheduling) estimates signed local defects by Richardson comparisons, transports them through the remaining discrete flow and a terminal-report adjoint, and optimizes the squared norm of their sum before a capped direct refinement. We show that this transported signed sum is the first-order terminal-report error and quantify, under an aligned-defect model, the inefficiency of diagonal criteria. Across six budgets on a regime-switching simulation and chronologically held-out Mag7 contexts, TAPES gives the lowest design- and audit-report errors in all 24 dataset--budget--report rows within the reported five-pipeline panel. Cross-budget mean relative audit-error reductions are about 10--11% on Regime-SV and about 1% on Mag7, with Bonferroni intervals excluding zero; all displayed winners persist under a doubled-resolution reference, and a small Regime-SV Heun probe preserves the TAPES-first audit ordering. TAPES changes only the offline schedule and retains the prescribed deployment NFE; claims concern numerical fidelity to frozen generators.","authors":["Yilie Huang"],"cats":["q-fin.CP","cs.LG"],"primary":"q-fin.CP","published":"2026-11-12T08:53:44Z","topic":3},{"id":"2611.16478","title":"One Model, Two Precisions: Does Disagreement Between Quantized Copies Tell You When a Small Model Is Wrong?","summary":"Local language models often exist at several precisions, and quantization is known to change individual answers. This study asks whether the disagreement between two precisions of the same weights, an 8-bit copy that answers and a lower-precision copy that reads the same prompt, works as a per-question error score for abstention, escalation and re-asking. The cost is one extra forward pass over fewer bytes, with no sampling and no training. We simulate round-to-nearest quantization with per-group scales (3 to 6 bits, groups of 32 to 256) on Qwen3-0.6B and Qwen3-1.7B and score 8,398 multiple-choice questions from ARC-Easy, ARC-Challenge, MMLU and HellaSwag subsets. Error detection is measured by AUROC, risk-coverage curves and ECE on a held-out half, with paired bootstrap intervals over questions. For Qwen3-0.6B, the entropy of the 8-bit answer reaches AUROC 0.682; the best disagreement score, selected on a development half, reaches 0.684 (+0.001 [-0.009, +0.011]), and combining it with the confidence features adds +0.009 [+0.002, +0.015], mostly on ARC. On Qwen3-1.7B the selected score is -0.001 [-0.009, +0.006] from entropy and the combination +0.003 [-0.001, +0.006] on the test half. Gaussian weight noise at matched error energy is about as informative as rounding (0.6B only), so we find no evidence that the signal is specific to quantization, and shuffling the features within task and entropy cells removes the gain. One extra 8-bit pass that shifts the option order does better (+0.049 [+0.031, +0.067] over entropy) at about a third more weight traffic; routing to a larger model on free entropy beats both extra-pass scores at equal total traffic. After logistic recalibration the 0.6B confidences are calibrated within a few points of ECE. A deployment recipe and the cases where the method does not help are given.","authors":["Mohammed Nayeem Khan","Allan Kevin T"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T08:47:29Z","topic":4},{"id":"2611.16477","title":"Where in the Input Does a One-Pass Decision Encoder Look?","summary":"Laya is a small model that answers a multiple-choice question about a text in one pass: the question, the options (each with a name and a description) and the text form one sequence of at most 512 tokens, the window, in the English release. A longer text is silently cut at the end, and we measure what that costs. Test documents place a held-out row's text, the evidence, at a chosen token offset among distractor text; 177 items, five tasks, three checkpoints (released, short-row fine-tune, long-document fine-tune), four window sizes and 93,275 scored requests. When the evidence lies beyond the window, accuracy falls to 26.4% for the released checkpoint (27.2% for both fine-tunes), near the 27.0% of guessing and below the 35.6% of always giving the commonest label. Confidence gives no warning (AUROC 0.50 to 0.55 for telling cut from read requests; 0.5 = chance), and the runtime's truncation flag also fires on requests whose evidence was visible. The released checkpoint reads visible evidence badly once text precedes it: 38.5% pooled against 60% when the evidence comes first, so a larger window alone gains +6.2 points. Fine-tuning on long documents adds +27.6 points on visible evidence over the short-row fine-tune (95% interval +22.3 to +33.1) and nothing on cut evidence. That is an in-family result: training rows and test documents share generator and question types, and the main filler is of the training kind. With encyclopedic filler never seen in training the gain is +13.2 (+9.9 to +16.6), with rows of other tasks as filler -2.0. Among training-free rearrangements a 312-token window scan helps most (+11.9 points over the default for the released checkpoint) at six to eleven forward passes. Documents are synthetic, labels are LLM-made and one backbone is studied, so these diagnose this model, not long documents in general.","authors":["Mohammed Nayeem Khan","Allan Kevin T"],"cats":["cs.CL","cs.AI"],"primary":"cs.CL","published":"2026-11-12T08:42:11Z","topic":6},{"id":"2611.16473","title":"Robots That Talk Too Much and Don't Listen Enough","summary":"Until recently, the biggest obstacles to natural spoken interaction with social robots were speech recognition errors and turn-taking failures. Advances in speech recognition have enabled much better spoken interaction, but turn-taking failures and other misunderstandings continue to be far too prevalent. This short paper looks back at earlier methods for improving speech recognition and turn-taking in interaction with social robots, and then looks at what can be done today, arguing that multimodal bidirectional streaming is needed to reduce the frequency of turn-taking failures.","authors":["Graham Wilcock","Kristiina Jokinen"],"cats":["cs.RO","cs.AI","cs.HC"],"primary":"cs.RO","published":"2026-11-12T08:32:18Z","topic":2},{"id":"2611.16466","title":"How Should Planning Agents ReSHAPE Their Search Space?","summary":"Planning agents and benchmarks treat a user's initial specification as fixed. Yet the requests users write exclude plans they would gladly take, since they rarely discover the possible trade-offs they would accept until seeing the options. Agents already revise requests through interaction, but whether this helps planning is unknown. We formalize this step as the ReSHAPE task, which judges a revision by whether its search space matches the acceptable plans and a planner finds one there. Using ReSHAPE Planter, which builds instances from planning benchmarks with a human-validated simulated user, we analyze 5,800 agent-revised specifications in travel and shopping. We find that reshaping matters: one correct trade-off makes the same planner reach an acceptable plan up to 4.6x as often, and search space quality predicts what the planner finds. Current agents, however, admit at most 32% of the acceptable plans, where that trade-off alone admits over 52%, because they ask about preferences instead of showing plans; in travel, showing plans raises F1 by 9.7 points over asking, with two model families. Planning agents should thus treat a request as a hypothesis to test against the user's reactions to concrete plans to recognize user trade-offs.","authors":["Yilin Zhang","Xinran Zhao","Tong Chen","Ziming Yuan","Tongshuang Wu"],"cats":["cs.AI","cs.CL"],"primary":"cs.AI","published":"2026-11-12T08:22:18Z","topic":0},{"id":"2611.16461","title":"Helgason-Fourier Random Features for Scalable Hyperbolic Latent Variables","summary":"Hyperbolic geometry is well suited to embedding hierarchical data, capturing tree-like structure faithfully even in the low dimensions where Euclidean dimensionality reduction is severely constrained. We introduce Random Helgason--Fourier Features (RHFF), an explicit finite-dimensional feature map for kernels on hyperbolic space. We characterize the spectral density from which RHFF frequencies are drawn, allowing kernels such as the Matérn family to be specified directly in the frequency domain. We also prove a uniform approximation error bound for RHFF. Building on this representation, we develop a hyperbolic latent variable model called RHFF-LVM, which requires no numerical quadrature, variational bounds or auxiliary variables unlike previous hyperbolic GP-LVMs. Our RHFF-LVM recovers the metric of a synthetic tree more faithfully than all baselines and attains the best mean rank, driven by large gains on tree-structured atlases at $Q\\!=\\!2$. Furthermore, it fits $5\\times10^{5}$ points in $505$ seconds, making it the first non-parametric hyperbolic LVM to reach this scale.","authors":["Koshi Watanabe","Daichi Mochihashi","Hideitsu Hino","Takahiro Ogawa","Miki Haseyama"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T08:13:11Z","topic":7},{"id":"2611.16460","title":"Fine-Tuned Small Language Model for Synthetic Clinical Data Generation and Predictive Healthcare Analytics","summary":"The reduced availability of medical datasets represents an important challenge in the development of predictive models for remote healthcare solutions, especially when the available data are imbalanced or difficult to extend through traditional methods. Small Language Models (SLMs) have become a promising alternative to Large Language Models (LLMs) for addressing these limitations due to their lower computational requirements and adaptability to specialized tasks. However, their performance remains highly dependent on the quality and representativeness of the available training data. In this paper, we propose a method for clinical data augmentation based on a fine-tuned Microsoft Phi-2 model that is used to generate synthetic instances, with the purpose of improving deep learning-based model for predicting remotely monitored vital signs. The Phi-2 model was adapted to the medical domain through supervised fine-tuning on the MedQuad dataset, through a Quantized Low-Rank Adaptation (QLoRA)-based method and by applying few-shot prompting techniques. The resulting model was used afterwards in the augmentation of a small-sized dataset, composed of vital sign values and labels associated with pathological and physiological states. The experimental results demonstrate that fine-tuned SLMs improved the performance on medical domain, achieving higher scores for both lexical (ROUGE) and semantic (BERTScore) evaluation metrics compared to the pretrained model. Moreover, the fine-tuned SLM is effective for generating structured synthetic clinical data. When used for data augmentation, the generated dataset improved the performance of a LSTM predictive model, increasing the classification accuracy from 89.0% to 93.7% and showing consistent improvements in precision, recall, and F1-score.","authors":["Mihnea Sebastian Burlacu","Ionut Anghel","Tudor Cioara"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T08:12:17Z","topic":6},{"id":"2611.16455","title":"Rational Alignments of Superintelligences via Debates","summary":"Artificial superintelligences (ASI) is artificial intelligences with greater capability. However, as human principal has to use it, we have to ensure that the agent will perform in the interest of the human. We investigate the framework of ASI-assisted question answering where ASI agents know the correct answer but the human does not. We provide some example of mechanism for rational alignment, where the agent utility is unchanged, but the interaction between the agent and the human principal is controlled to be in a certain fashion. We show specifically that a debate framework where the agents can further reason and behave in a subgame perfect equilibrium manner will lead to a more predictable result, and in some case will be favorable.","authors":["Thanawat Sornwanee"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T08:07:48Z","topic":1},{"id":"2611.16450","title":"A Recipe for Physical Action Model Training","summary":"Choosing robot model capacity and demonstration budgets also requires decisions about data composition, representation, optimization, and deployment. We connect these decisions through a common conditional-distribution target. Nested learning oracles give exact risk accounting; local information geometry, conditional mutual information, and empirical convergence provide conditional training prescriptions. Known-target numerical checks expose capacity-data coupling, representation-induced mixture mismatch, and an example where deployment KL increases while negative log-likelihood decreases. An analysis of 206 public PushT trajectories evaluates fixed-feature information-design criteria without fitting a policy. Under asymmetric evaluation targets, the selected allocations reduce this criterion by 4.5-6.2% relative to target matching; balanced targets show no benefit. These results clarify which quantities can support training recommendations and when an apparently favorable predictive metric has a different interpretation.","authors":["Shaojiang Wang","Geer Yang"],"cats":["cs.LG"],"primary":"cs.LG","published":"2026-11-12T08:03:30Z","topic":2},{"id":"2611.16445","title":"FPSRG: Heterogeneous User Simulation First-Person Post-Training and Multidimensional Semantic Response Grids","summary":"Simulating heterogeneous individuals with large language models is an active research direction, and heterogeneous user simulation requires models with different characteristics. We study profile construction: given dimensions, items, and scoring rules, build model configurations that can be retrieved by multidimensional target scores. We introduce First-Person Semantic Response Grids (FPSRG). Shared post-training first teaches the model to judge and answer as the participant; self-report and behavioral targets then refine dimension-specific control vectors. With the model frozen, we measure joint scores for vector-coefficient combinations, filter inadmissible configurations, and store the results in a searchable catalog. Per-token configuration routing, continuous batching, and isolated caches support parallel measurement of different profiles. Non-thinking five- and ten-dimensional Qwen instances and a thinking-mode six-dimensional instance following full-parameter post-training of Huihui demonstrate the complete workflow. With matched data, layers, and norms, refinement increases mean five-axis composite control span from 30.75 to 40.45. Within the five-dimensional 25-75 target region, the qualified five-level vector catalog covers 92.2% of fixed targets under per-axis error at most 20 and total error at most 60; combining it with the prompt catalog increases coverage to 98.7%. At a fixed coefficient range, three-, five-, and nine-level catalogs achieve 71.2%, 92.2%, and 94.6% coverage, demonstrating the gains from denser measurement. These results describe measured reachable regions under a fixed assessment protocol. Through shared adaptation, a small set of learned directions, and reusable joint measurements, FPSRG constructs heterogeneous individual configurations without retraining a model for every profile.","authors":["Ruimin Pei","Haibo Liu","Fang Zhang","Xunhua Guo"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T07:58:50Z","topic":6},{"id":"2611.16442","title":"DesignTaste: Benchmarking AI-Generator Visual Design Against Award-Wining Works","summary":"Text-to-image benchmarks commonly assess text-image alignment or overall preference. Whether models can produce good designs when a brief leaves the appearance open remains less well understood. We introduce DesignTaste, a visual design benchmark that compares generated images with award-winning works. From these works and their publicly available context, we derive 403 design briefs: 100 poster tasks from 100 Beste Plakate 2025 and 303 tasks spanning 20 categories from ADC 2026. Six leading commercial text-to-image models each generate an image from the same brief. A single model judge scores each output and the corresponding award-winning work independently on five evaluation dimensions. On each dataset, only ArtArk achieves a higher mean overall score than the award-winning works. All six models score lower in Creative Originality on both datasets. These results show that producing finished, visually appealing images does not imply that a model can develop good design solutions.","authors":["Zhou Chen","JiaXiang Li","DanYing Pan","Ye Li","TianRui Yang","EnBao Zhang","YaNan Xiao","LiXin Sun","Anqi Liu","Dong Zhao","JianYuan Li","YuanYuan Ding","Kai Kong","Xinwen Xu","HuiTang Du","Min Zhou","Zhiqiang Li"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T07:52:27Z","topic":3},{"id":"2611.16433","title":"daVinci-Prover: Agent-Native Mid-Training for Formal Theorem Proving","summary":"Advances in large language models (LLMs) are rapidly transforming formal theorem proving. Yet further progress remains constrained by the lack of a transparent and effective data recipe for systematically improving theorem-proving models. In this work, we argue that effective agentic provers require strong mathematical skill priors to ground theorem proving and agentic trajectories to enable long-horizon, interactive proof construction. To this end, we synthesize an agent-native 33.7B-token mid-training corpus covering decomposition, search, self-verification, and revision, together with verifier-grounded agentic trajectories. We further introduce the daVinci-Prover harness, which enables models to construct proofs and interact with a Lean verifier in an isolated workspace, and use it to collect approximately 36K high-quality agent trajectories for supervised fine-tuning. Our mid-training recipe yields an average improvement of 13.7 points over the matched Qwen3-8B SFT-only baseline across MiniF2F, FATE-M, and ProofNet. When applied to the Qwen3.5 series, our approach also yields strong performance across theorem-proving benchmarks and achieves near-saturating MiniF2F performance with memory compaction. The results highlight agent-native mid-training as an effective foundation for developing stronger interactive theorem-proving agents.","authors":["Wenhong Zhu","Xuefeng Li","Jiashuo Zhang","Yu Chen","Tiantian Mi","Qipeng Guo","Pengfei Liu"],"cats":["cs.CL"],"primary":"cs.CL","published":"2026-11-12T07:44:09Z","topic":0},{"id":"2611.16426","title":"SoScale: Second-Order Scale Search for FP4 Quantization","summary":"Low-bit quantization reduces the memory footprint and inference cost of large language models (LLMs), yet accurate FP4 quantization remains challenging. Practical formats such as MXFP4, NVFP4, and HiF4 pair a 4-bit floating-point payload with block-wise or hierarchical scale codes, making accuracy highly sensitive to how each local distribution is scaled. Default maximum-absolute-value scaling preserves dynamic range, but can spend scarce FP4 levels on rarely occupied outlier regions and leave dense tensor values poorly resolved. Existing scale-search methods minimize local element-wise reconstruction error, optimizing tensor fidelity rather than layer-output fidelity. We propose SoScale, a second-order weight scale-code search method for hierarchical FP4 quantization. SoScale searches hardware-valid scale codes under a Hessian-aware layer reconstruction objective and couples this search with GPTQ-style error recovery, selecting scales by their impact on layer outputs. Across Qwen3 models, SoScale consistently reduces downstream quantization degradation under MXFP4, NVFP4, and HiF4.","authors":["Pengxiang Zhao","Xing Li","Hui-Ling Zhen","Weizhe Lin","Xianzhi Yu","Zhenhua Dong"],"cats":["cs.LG","cs.AI"],"primary":"cs.LG","published":"2026-11-12T07:36:46Z","topic":4},{"id":"2611.16425","title":"RMIT-ADM+S at TREC RAG 2026: Research Agents, Retriever Ablations, and Open-Weight Replication","summary":"We describe the RMIT-ADM+S team's participation in the TREC 2026 Retrieval-Augmented Generation (RAG) track using the ClimbMix corpus. We submitted seven research-agent runs using proprietary OpenAI and open-weight models, exploring different retrieval strategies, writer models, and agent workflows. Pre-submission ablations identified writer selection and adjacent-chunk expansion as the main factors affecting answer quality. In the official NIST evaluation, six of seven runs exceeded the track median across all three rubric/nugget collections. Our best-performing run, brief-hybrid, achieved the highest all-item coverage among our submissions and tied the track maximum (0.892) for Batch-1 vital coverage. Human side-by-side preferences differed across evaluation batches: the open-weight owa-glm5 led Batch 1 despite lower automated coverage, while sol-research-first led Batch 2. These findings highlight differences between automated coverage measures and human preferences, as well as the potential of lower-cost open-weight models for research-agent systems.","authors":["Oleg Zendel","Kun Ran","Chenglong Ma","Nguyen Khoi Vo","Shuai Wang"],"cats":["cs.IR","cs.AI","cs.CL"],"primary":"cs.IR","published":"2026-11-12T07:34:28Z","topic":0},{"id":"2611.16424","title":"PhysCapBench: A Cross-Domain Benchmark for Physically Grounded Video Captioning","summary":"Physical AI aims to learn transferable physical knowledge from real-world experience to understand, predict, and act in the physical world. Internet videos and embodied demonstrations provide complementary observations of motion, interaction, and state evolution. A foundational data challenge is to measure and preserve this physical information as training supervision. In video-language pretraining, captions express observed processes in language, yet a semantically correct description can omit the interactions and transitions that define those processes. Rich physical experience in video does not automatically translate into sufficient language supervision. This gap calls for a yardstick to assess annotation quality, compare captioning strategies, and guide training-data refinement. Motivated by our data construction for world-model pretraining, we introduce PhysCapBench, to our knowledge the first unified physical-caption benchmark spanning all three domains: third-person scenes, egocentric manipulation, and handheld-gripper demonstrations. It provides human-verified annotations for 1,500 video sequences and evaluates accuracy and coverage through seven physical dimensions. We examine downstream utility through video reconstruction and manipulation-policy fine-tuning. Physical captions reduce reconstruction LPIPS by 8.6-9.5% on third-person videos and 26.9-30.9% on egocentric videos relative to metadata captions. Training-instruction enrichment improves both PhysCapBench scores and policy success, including a 7.5-percentage-point gain for pi0.5 on a RoboTwin 2.0 subtask. PhysCapBench makes the physical content of language supervision inspectable and comparable, providing a cross-domain basis for its evaluation and improvement.","authors":["Haofeng Wang","Pengcheng Wen","Ruiqi Li","Yuhao Zhang","Tiange Zhang","Boren Zheng","Xuanyi Liu","Siyu Tao","Danni Yu","Gan Lin","Haocheng Tang","Tianzizhe Guo","Mingyuan Su","Boyuan Chen","Zhongyuan Wang"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T07:33:09Z","topic":3},{"id":"2611.16422","title":"BlinkMimic: Training a Humanoid Motion Tracker in Minutes on a Single GPU","summary":"Whole-body motion tracking lets a humanoid robot follow reference motions, but training a tracking policy for a library of motions has usually taken hours of GPU time. SuperMimic showed that the policy can instead be trained by backpropagating through a world model learned from parallel simulation, and InstantTrack added Muon updates, a residual policy around the reference and an averaged deployed policy. We build on both and present BlinkMimic, a recipe that makes this training faster. On a simulated Unitree G1, BlinkMimic trains one policy for all 40 LAFAN1 clips in 3.4 minutes on a single RTX 4090, measured until three consecutive checkpoints reach a normalized return of at least 0.82 with no failed clip (mean of three seeds). We split training time into the cost of an iteration, the iterations needed to fit the motions, and the iterations needed to remove the last failures, and assign each component, inherited or new, to one of these terms. The new components are a step size that decreases once training has progressed, resets directed by failures of the deployed policy under deployment conditions, and GPU graph capture of the whole iteration. A leave-one-out ablation with three seeds per component shows which term each component affects, and further measurements of the gradient, the update rule and the failure cases explain these effects. Code and per-seed results are available at https://github.com/fuyichen0725/blinkmimic.","authors":["Fuyi Chen"],"cats":["cs.RO","cs.LG"],"primary":"cs.RO","published":"2026-11-12T07:31:42Z","topic":2},{"id":"2611.16414","title":"Small Language Models for Inference-Capability Routing: Effects of Model Capacity and Prompt Optimization on Downstream Quality and Efficiency","summary":"Optional reasoning, retrieval, web search, and tool use can improve large language model responses but incur unnecessary computation when activated indiscriminately. This study investigates whether prompt optimization can compensate for limited router capacity and whether classification gains translate into downstream efficiency. The proposed framework uses prompt-driven small language models to make four independent capability decisions, allowing multiple capabilities to be selected per query. Qwen2.5 routers ranging from 1.5B to 14B parameters are compared across five prompting strategies, followed by an evaluation of capability-specific optimized configurations using a fixed GPT-OSS-120B downstream model on 127 questions. Under minimal prompting, increasing router capacity does not consistently improve classification performance. Prompt optimization narrows the capacity gap for reasoning and calculator routing, whereas RAG and web search retain larger benefits from higher capacity. Similar classification F1 scores do not necessarily imply equivalent downstream behavior: false-positive activations increase unnecessary computation, while missed capabilities can reduce answer quality. The 14B router achieves a mean BERTScore F1 of 0.8408, compared with 0.8418 when all capabilities are enabled, while reducing average downstream token consumption by 68.7\\% (8,837 to 2,764 tokens) and median downstream latency by 80.5\\% (9.79 to 1.91 seconds). These savings exclude router overhead, which is measured separately in a different inference environment and increases with router capacity. The findings demonstrate that the relationship between router capacity, prompt design, and downstream efficiency is capability-dependent. Consequently, router selection must consider error profiles and routing overhead alongside classification performance.","authors":["Sailendra Kolluru","Maheshkumar Duvvarapu"],"cats":["cs.AI"],"primary":"cs.AI","published":"2026-11-12T07:25:46Z","topic":4},{"id":"2611.16409","title":"Experience-Handoff Reinforcement Learning for Recursive Self-Improvement","summary":"It is common to train Large Language Models (LLMs) for agentic tasks with Reinforcement Learning (RL) from scalar rewards. However, this throws away the fine-grained information about the environment that the agent experienced during the trial (e.g., error logs). To utilize this rich information, recent works wrap the agent in a harness that refines its own inputs through reflection or compaction and hands past experience to future rollouts, but the construction of the handed-off artifacts (e.g., a playbook) is itself not learned. We want the best of both worlds: gradient updates from a scalar reward, while also utilizing environment information to hand off experience. As a concrete testbed, we fix the Agentic Context Engineering (\\ace{}) harness and train one model to act, reflect, and curate task-local playbooks. Our method, \\acerl{}, evaluates candidate experience-handoff artifacts by their downstream effect on future actor rollouts, and co-optimizes the role-specific behaviors (action generator, insight reflector, and playbook curator) under a fixed scaffold. Across agentic benchmarks, \\acerl{} achieves higher held-out performance than baseline methods. The results suggest that agent training should optimize not only the actions in the current rollout, but also the model-written experience that future rollouts inherit.","authors":["Doyoung Kim","Byeongguk Jeon","Kailin Zhang","Sherry Yang"],"cats":["cs.LG","cs.CL"],"primary":"cs.LG","published":"2026-11-12T07:18:30Z","topic":2},{"id":"2611.16399","title":"Saidzoda-Kazakh-EmbeddingGemma-300M: How Far One Epoch of Generated Questions Takes a Small Kazakh Embedder","summary":"EmbeddingGemma-300M is a small open embedder, and used as released it finds the right passage first for 60.8% of the Kazakh Belebele retrieval questions. We present Saidzoda-Kazakh-EmbeddingGemma-300M, the same model fully fine-tuned for one epoch on 208,511 pairs: 166,263 Kazakh questions that our own language model wrote about source passages from encyclopedias, Wikipedia, news and books, each checked by a second model, plus English and Russian replay. On Kazakh Belebele, recall@1 rises from 60.8% to 78.0% and recall@10 from 76.3% to 91.4%. Our model scores above 161 of the 194 models with a published MTEB score on the same test, among them OpenAI text-embedding-3-large (47.1%). The other 33 score higher: the ten best are large multi-billion-parameter or closed models, reaching up to 92.2%, and at least 9 are open models of up to 0.6B parameters, the smallest with 339M. We also scored 6 Kazakh-specific embedders shared on Hugging Face with the same code in one run: our model (77.3% there) is ahead of 4 of them, including the two E5-base models of similar size (72.6% and 69.0%), and behind kaz-embedder (86.0%, three towers with 1,695M parameters in total) and a BGE-M3 fine-tune (81.0%); at recall@10 its interval overlaps those of its nearest neighbours on both sides. Our four-language model, trained with a capped share of the same Kazakh pairs, is 2.8 points lower at recall@1, with overlapping recall@10 intervals. The Kazakh model also scores higher than the released model on the Uzbek and Kyrgyz versions of Belebele, but our Tajik model, trained without any Turkic pairs, gains as much there, so these gains are not specific to Kazakh training. Our Kazakh Wikipedia test questions come from the same generator family as the training questions, so we rest the main claims on Belebele.","authors":["Tohir Saidzoda","Madina Abdusalomzoda","Abubakr Abdullozoda","Amin Jalilov","Faroiz Bobokhon","Aibek Shokanuly"],"cats":["cs.CL"],"primary":"cs.CL","published":"2026-11-12T07:09:08Z","topic":6},{"id":"2611.16393","title":"TransXplainRRG: A Radiologist-Evaluated LVLM for Chest Radiograph Report Generation","summary":"Recent advancements in Large Vision-Language Models (LVLM) enhance X-ray report generation, but often neglect clinical context, overlook workflow integration, and lack explainability, which hinders clinical adoption due to interpretability and reliability concerns. This paper addresses the dual challenges of performance and expert explainability in automated X-ray report generation for clinical AI applications. Our proposed approach, TransXplainRRG, utilises an off-the-shelf Swin Transformer model along with a transformer-based text encoder that incorporates patient medical history to generate a radiology report. Further, we explore an expert-guided `inside-out' approach and extract only abnormal findings for radiology report refinement. Thus, this study bridges the gap between high-performance automation and interpretability critical for clinical practice by combining state-of-the-art transformer-based vision encoders, text encoders, and LLMs. TransXplainRRG is trained and tested on the large-scale MIMIC-CXR dataset and further evaluated on the unseen IU X-Ray dataset to demonstrate its generalizability and robustness. We employed the GREEN metric as a complementary measure of clinical consistency, achieving scores of 0.327 on MIMIC-CXR and 0.605 on IU X-Ray, both higher than those achieved by MAIRA-2 under the same evaluation setting. It also performed competitively across standard evaluation metrics, frequently matching or surpassing state-of-the-art models. Additionally, we introduce qualitative evaluation metrics developed from radiologists' viewpoints and conduct a preliminary radiologist evaluation on 100 generated reports to assess their clinical relevance. Quantitative and qualitative analyses further demonstrate the potential of our TransXplainRRG framework for reliable and explainable radiology report generation.","authors":["Chayan Mondal","Duc-Son Pham","Tele Tan","Ashu Gupta","Vincent T. Y. Tiong","Shreya Ghosh","Tom Gedeon"],"cats":["cs.CV","cs.AI"],"primary":"cs.CV","published":"2026-11-12T07:05:39Z","topic":1},{"id":"2611.16384","title":"Beyond Compressing Everything: Contrastive Memory Learning for Selective Context Compression","summary":"As large language models (LLMs) advance toward increasingly long inputs, soft context compression offers a promising approach to efficient inference by converting long inputs into compact memory slots. Under limited memory budgets, effective context compression remains challenging because task-relevant evidence varies substantially across queries, ranging from localized evidence to multiple distant spans or information dispersed throughout the full document. Instead of compressing everything indiscriminately, we propose CoMem, a query-guided selective context compression framework that learns memory representations conditioned on both the document and the query. CoMem develops Contrastive Compressor Learning to achieve query-sensitive yet context-preserving memory representations. To inject a strong prior over query-dependent evidence distributions, we devise a same-context hard negative mining strategy by pairing a single document with multiple queries, requiring the compressor to selectively retain information according to each query's distinct evidence needs. To address the restricted information flow imposed by causal masking, we further enable bidirectional attention among memory slots. Experiments demonstrate that CoMem consistently outperforms existing context compression methods across compression ratios on question answering and RULER-NIAH benchmarks. Our code is available at https://github.com/scar-on/CoMem.","authors":["Sikui Zhang","Haowei Liu","Ziyun Gan","Chunfeng Yuan","Guangze Gao","Nian Liu","Boyu Cai","Bo Li","Xiaolei Lv","Jun Gao","Bing Li","Weiming Hu"],"cats":["cs.CL","cs.AI"],"primary":"cs.CL","published":"2026-11-12T06:51:49Z","topic":4},{"id":"2611.16378","title":"Evidence-Governed Adoption of Self-Generated Updates in Autonomous AI Agents","summary":"Autonomous agents modify prompts, memories, tools, workflows, and policies using interaction histories. Yet a plausible modification is not necessarily suitable for persistent adoption: candidate selection is adaptive, evaluators may be imperfect, and local gains can hide degradation against an earlier baseline. We introduce Evidence-Governed Evolution (EGE), a framework for governing adoption of self-generated agent modifications. EGE separates candidate generation from confirmatory adoption, evaluates frozen candidates on fresh paired audits, requires meaningful improvement on a prespecified primary objective, protects declared capabilities through reference-relative non-inferiority tests, and applies governance constraints before bounded canary deployment. We establish an evaluator-calibrated simultaneous lower-bound construction for a countable sequence of adaptively generated candidates. It combines candidate-wise Hoeffding bounds with independent calibration samples comparing the operational evaluator with a prespecified reference procedure. We also formalise reference replacement through bridge tests, derive a cumulative degradation bound along a versioned reference lineage, and provide an audit-planning condition relating effect-size margin, sample sizes, and acceptance probability. A Monte Carlo stress test illustrates calibration under candidate-dependent evaluator optimism; it is a stylised statistical experiment, not an evaluation of a deployed agent. The guarantees are conditional on bounded paired scores, fresh sampling, a valid reference-evaluation construct, and declared evaluation distributions; they do not establish universal agent safety or robustness to arbitrary distribution shift.","authors":["Subodh Lonkar"],"cats":["cs.AI","cs.MA"],"primary":"cs.AI","published":"2026-11-12T06:45:02Z","topic":0},{"id":"2611.16375","title":"Recursive State-Augmented Language Models: Persistent Latent-State Adaptation of Pretrained Language Models","summary":"Language models rely on finite token contexts, limiting access to earlier interactions. We introduce Recursive State-Augmented Language Models (Recursive SALM), a framework in which a persistent latent state evolves with each observation and generates adaptations of a pretrained language model. A modular read-transition-write architecture separates observation encoding, latent dynamics, and predictor adaptation. We develop deterministic and stochastic formulations. The experiments evaluate only deterministic SALM with frozen GPT-2 small and exclude earlier evidence from its decoder context. On a four-turn hypothesis-inference task, latent-gated LoRA reaches 89.1% accuracy, comparable to 88.0% for a trained full-context LoRA baseline, with earlier evidence available solely through a 64-dimensional state. Zeroing or shuffling the state substantially reduces accuracy. In a two-turn task with codewords unseen in training, several interfaces improve candidate ranking over chance, but rarely generate the exact codeword under the evaluated training schedule. These experiments show that persistent latent-conditioned adaptation can support sequential inference when earlier evidence is absent from the decoder context.","authors":["Omer Haq"],"cats":["cs.LG","cs.AI","cs.CL","stat.ML"],"primary":"cs.LG","published":"2026-11-12T06:42:54Z","topic":6},{"id":"2611.16369","title":"Bait: Mitigating Harmful Agentic Reinforcement Learning via Probing Reward Function","summary":"We expose a safety vulnerability on nascent agentic reinforcement-learning-as-a-service. Specifically, with a cost as low as \\$20, we show that safety alignment of a large language model (e.g., Inkling) can be completely subverted through rounds of harmful agentic reinforcement learning (agentic RL) with a toxic environment from its users. We also demonstrate that upon a successful harmful agentic RL attack via the service API, the attackers can potentially use the safety-degraded model to assist with real-world harmful activities, e.g., exploit software vulnerabilities and launch code re-use attacks. To mitigate harmful agentic RL, we propose Bait, a defense that mitigates the attack via proactively probing the reward function. Experimental results show that such a simple defense can sufficiently mitigate learning from a harmful reward function and can still learn from a benign reward function that promotes benign agentic tasks. Further analysis on both empirical and theoretical aspects shows that an adaptive attacker who assumes the knowledge of Bait defense cannot perfectly subvert it due to a systematic challenge. Source code is available at https://github.com/git-disl/Bait.","authors":["Tiansheng Huang","Ananth Balashankar","Virat Shejwalkar","Santosh Pande","Ling Liu"],"cats":["cs.LG","cs.CR"],"primary":"cs.LG","published":"2026-11-12T06:34:34Z","topic":0},{"id":"2611.16368","title":"ReliaGuard: Reconstruction-Aware Watermarking for Source Attribution and Blind Edit Localization","summary":"Generative image editing facilitates content creation but complicates source attribution and edit verification. Existing watermarking methods combine copyright protection with tamper localization, yet background reconstruction can weaken spatial evidence even when source identifiers remain recoverable. We propose ReliaGuard, a reconstruction-aware watermarking framework for source attribution and blind edit localization. The writer optimizes keyed feature responses in both the protected image and its variational autoencoder reconstruction under image-distortion and source-preservation constraints. Using only the received image and a secret key, the reader combines watermark responses with local image context through residual evidence refinement and predicts an edit mask through native-resolution-guided multiscale reading. It is trained on complete editing outputs, locally spliced counterparts, and processed versions of both. On an independent test set of 300 source images and 900 editing cases from three image domains, our method achieves edit-region intersection-over-union scores of 85.61% for complete outputs and 94.90% for locally spliced outputs at a threshold calibrated to a nominal 1% background false-positive rate on complete outputs. The complete-output result exceeds the strongest baseline by 45.02 percentage points, while decoded identifier-bit accuracy reaches 100% in both conditions. Across seven families of image processing, the corresponding localization scores are 81.01% and 86.49%, with identifier-bit accuracies of 92.76% and 92.79%, respectively. These results support source recovery and blind edit localization under background reconstruction and subsequent image processing.","authors":["Hailong Wang","Wuzheng Tan","Zixin Zhang","Xujun Fang","Fengjuan Gong"],"cats":["cs.CV","cs.AI","cs.CR"],"primary":"cs.CV","published":"2026-11-12T06:33:37Z","topic":1},{"id":"2611.16361","title":"MutAtlas: A PDB-Wide Energy-Guided Atlas of Protein Mutation Effects","summary":"Protein mutation effect prediction is fundamental to protein engineering and disease variant interpretation, yet experimentally measured mutation data remain accurate but extremely sparse. To provide scalable supplementary mutation signals, we construct a PDB-wide mutation augmentation dataset that exhaustively enumerates single-site substitutions on experimentally resolved protein structures and aligns mutation signals from physics-based energy models, protein language models, and inverse folding models. Large-scale analysis under a unified mutation preference representation reveals substantial differences in the consistency, concentration, and substitution patterns of mutation distributions across models, indicating that disagreement is pervasive and reflects conflicting inductive biases rather than random noise. Motivated by these observations, we propose an unsupervised multi-source mutation preference distillation framework that learns from relative mutation preferences while explicitly modeling cross-source disagreement. Without using any experimental mutation labels during training, our approach achieves the best overall performance among the evaluated zero-shot baselines and naive multi-source fusion strategies on ProteinGym. We release the dataset and evaluation pipeline to support reproducible studies of protein mutation effects.","authors":["Ruihan Guo","Chaoran Cheng","Zhanghan Ni","Neil He","Bangji Yang","Ge Liu"],"cats":["cs.LG","q-bio.GN"],"primary":"cs.LG","published":"2026-11-12T06:22:47Z","topic":4},{"id":"2611.16359","title":"IsharaBART: An efficient BART-Based Framework for Low-Resource Bangla Sign Language Translation","summary":"Millions of Bangla Sign Language (BdSL) users lack reliable automatic translation tools. Existing systems are limited by data scarcity, unidirectional design, and computationally expensive architectures that are not suitable for low-resource deployment. We introduce IsharaBART, a bidirectional generative framework for gloss-sentence translation in BdSL. IsharaBART uses Bidirectional and Auto-Regressive Transformers (BART)-based conditional language modeling with denoising learning objectives. We developed a parallel dataset of 1,000 professionally annotated gloss-sentence pairs, balanced across tenses and structures. We expanded the dataset to 4,831 pairs through four augmentation techniques: rule-based tense transformations, masked token substitutions, back-translation, and one-shot prompting. A compact denoising sequence-to-sequence model was pre-trained from scratch on monolingual Bangla text from six different corpora, then fine-tuned on the annotated and augmented datasets. We compare IsharaBART against four larger baseline models under identical conditions, where it achieves comparable or superior bidirectional performance. On the gloss-to-sentence (G2S) task, it attains Bilingual Evaluation Understudy (BLEU-4) score up to 51.7, and on the sentence-to-gloss (S2G) task, it achieves 61.40, outperforming multilingual models despite having fewer than 10% of the parameters. Ablation studies demonstrate that pre-training provides 62.5-90.4% faster convergence and 11.5-118% higher BLEU-4 than scratch. Low-rank adaptation (LoRA) reduces trainable parameters by 99.3%, but with 28.0-52.2% loss of BLEU-4 compared to full-parameter fine-tuning.","authors":["Aonmoy Das","Ananna Dev Aishi","Md. Fazlul Kader"],"cats":["cs.CL"],"primary":"cs.CL","published":"2026-11-12T06:21:34Z","topic":4}],"days":[{"date":"2026-10-13","total":588,"counts":[106,53,59,76,59,18,176,41],"est":true},{"date":"2026-10-14","total":541,"counts":[87,76,43,119,70,38,60,49],"est":true},{"date":"2026-10-15","total":541,"counts":[92,81,54,76,27,43,108,60],"est":true},{"date":"2026-10-16","total":413,"counts":[78,54,37,58,29,29,87,41],"est":true},{"date":"2026-10-17","total":192,"counts":[40,25,15,25,25,10,38,13],"est":true},{"date":"2026-10-18","total":209,"counts":[29,25,19,46,13,21,38,19],"est":true},{"date":"2026-10-19","total":518,"counts":[114,88,47,93,41,52,36,47],"est":true},{"date":"2026-10-20","total":465,"counts":[121,28,84,42,37,37,79,37],"est":true},{"date":"2026-10-21","total":401,"counts":[92,36,40,64,48,28,60,32],"est":true},{"date":"2026-10-22","total":410,"counts":[82,62,29,49,78,21,70,21],"est":true},{"date":"2026-10-23","total":357,"counts":[64,43,32,57,43,18,54,46],"est":true},{"date":"2026-10-24","total":175,"counts":[32,16,14,30,19,4,37,25],"est":true},{"date":"2026-10-25","total":168,"counts":[44,12,10,30,18,3,32,18],"est":true},{"date":"2026-10-26","total":473,"counts":[85,52,38,47,61,19,80,90],"est":true},{"date":"2026-10-27","total":446,"counts":[76,67,27,80,71,27,62,36],"est":true},{"date":"2026-10-28","total":249,"counts":[47,37,22,40,27,12,37,25],"est":true},{"date":"2026-10-29","total":443,"counts":[75,66,49,58,35,35,80,44],"est":true},{"date":"2026-10-30","total":355,"counts":[64,64,43,43,53,11,36,43],"est":true},{"date":"2026-10-31","total":264,"counts":[29,32,37,29,32,16,45,45],"est":true},{"date":"2026-11-01","total":269,"counts":[40,27,19,40,32,30,46,35],"est":true},{"date":"2026-11-02","total":482,"counts":[106,43,67,53,67,34,53,58],"est":true},{"date":"2026-11-03","total":425,"counts":[89,38,43,77,47,9,98,26],"est":true},{"date":"2026-11-04","total":369,"counts":[103,44,18,37,59,15,59,33],"est":true},{"date":"2026-11-05","total":388,"counts":[70,78,31,50,54,12,47,47],"est":true},{"date":"2026-11-06","total":372,"counts":[63,56,33,41,41,22,86,30],"est":true},{"date":"2026-11-07","total":251,"counts":[28,38,33,45,20,18,50,20],"est":true},{"date":"2026-11-08","total":201,"counts":[32,24,26,32,20,10,28,28],"est":true},{"date":"2026-11-09","total":431,"counts":[103,39,26,73,78,26,52,34],"est":true},{"date":"2026-11-10","total":353,"counts":[71,28,28,64,56,11,60,35],"est":true},{"date":"2026-11-11","total":283,"counts":[62,28,23,34,34,20,45,37],"est":true}],"attention":[{"id":"2610.14162","title":"Next-Concept Modeling: Fast, Parallel Coarse-to-Fine Text Generation","hn":"50422168","points":186,"comments":47},{"id":"2611.10866","title":"The Surprising Effectiveness of Deleting Weights in LLM Post-Training","hn":"50447363","points":30,"comments":0},{"id":"2611.09286","title":"Do the day's losing stocks still bounce in the last half hour? A registered test","hn":"50451588","points":6,"comments":0},{"id":"2610.22192","title":"Terminal-RSI: Recursive Self-Improving Terminal Agents","hn":"50423647","points":5,"comments":1},{"id":"2611.00230","title":"LHCb finds evidence for the decay B0 to mu+ mu-, with improved B0s measurement","hn":"50433975","points":5,"comments":0},{"id":"1806.02404","title":"Dissolving the Fermi Paradox","hn":"50414985","points":4,"comments":0},{"id":"2101.09170","title":"Why are there Seven Sisters?","hn":"50446417","points":3,"comments":1},{"id":"2609.30468","title":"SR-Gadgets: Make Scan-Resistant Caching Practical","hn":"50401905","points":3,"comments":0},{"id":"2001.02209","title":"Correctness of Automatic Differentiation via Diffeologies and Categorical Gluing","hn":"50416442","points":2,"comments":0},{"id":"2508.05276","title":"An Overview of 7726 User Reports:Uncovering SMS Scams and Scammer Strategies","hn":"50457897","points":2,"comments":0},{"id":"2510.26787","title":"Remote Labor Index: Measuring AI Automation of Remote Work","hn":"50465606","points":2,"comments":0},{"id":"2607.21498","title":"Artificial Epanorthosis: Why LLMs overuse a classical rhetorical figure","hn":"50407306","points":2,"comments":0},{"id":"2608.00325","title":"Triton for MTIA: Bridging the Programming Model Gaps for Custom AI Accelerators","hn":"50431683","points":2,"comments":0},{"id":"2610.20344","title":"Doc2Harness: Benchmarking how well AI builds agent harnesses","hn":"50445107","points":2,"comments":0},{"id":"2610.20666","title":"ArXiv: Schelling-Point Arbitration as a Composable Solana Primitive","hn":"50451839","points":2,"comments":0},{"id":"2610.36669","title":"Proof Repair and (Co)algebra Equivalences","hn":"50436263","points":2,"comments":0},{"id":"2610.37407","title":"Constraining the Commit Boundary: DIO/ODI Authorization Guardrail for Agentic ER","hn":"50401134","points":2,"comments":0},{"id":"2611.01463","title":"DIMWIT: Directly Invoking Models Without Intensive Tuning","hn":"50457293","points":2,"comments":0},{"id":"2611.05173","title":"Correctness Firewalls for Synthesized Database Components","hn":"50456920","points":2,"comments":0},{"id":"2611.07672","title":"Non-axisymmetric MHD equilibria with an arbitrary number of field periods","hn":"50449349","points":2,"comments":0}]}