Research

• General in-context learning for embodied agents1 paper raises it · 0 address it

  Enables models to adapt to new tasks and environments without retraining, improving robustness in real-world deployment.

  • Has GPT-6-Astra Achieved Embodied Intelligence?raises
• Cost and latency in embodied control loops1 paper raises it · 0 address it

  High computational and time costs limit real-time applicability of agentic systems in dynamic environments.

  • Has GPT-6-Astra Achieved Embodied Intelligence?raises
• Physical-state verification and failure recovery1 paper raises it · 0 address it

  Agents must reliably detect and correct errors in physical interactions to avoid irreversible failures in real-world tasks.

  • Has GPT-6-Astra Achieved Embodied Intelligence?raises
• Long-horizon task preservation in embodied agents1 paper raises it · 0 address it

  Maintaining progress over extended sequences is critical for complex, multi-step tasks in robotics and navigation.

  • Has GPT-6-Astra Achieved Embodied Intelligence?raises
• Open-source stacks for generalist computer-use agents1 paper raises it · 1 addresses it

  Closed systems hinder reproducibility and systematic study of agent design for real-world computer tasks.

  • OpenGCUA: Toward a Fully Open Stack for Generalist Computer Useraises · addresses

Named by an AI from these papers' abstracts.

• Code models22+1000.0%
• Small LLMs11+1000.0%
• AI safety43+975.0%
• Training & fine-tuning58+866.7%
• Robotics28+600.0%
• Image & video generation34+580.0%
• Evals94+571.4%
• Computer vision43+437.5%

• arXiv cs.LGOct 17

  Sample-Efficient Reinforcement Learning with General Utilities via Neural Dual Critics

  Ankur Naskar, Vaneet Aggarwal

  Reinforcement learning with general utilities optimizes nonlinear functionals of policy-induced occupancy measures. We develop a natural actor-critic method for concave RLGU objectives admitting an exact variational dual representation, with explicit constructions for a broad class of separable occupancy utilities. Rather than estimating the occupancy measure and then forming the policy-gradient pseudo-reward, our method learns this pseudo-reward directly as a function of the state-action pair using a neural critic. The critic is trained by projected stochastic updates, while multilevel Monte Carlo combines outputs across training horizons to control its bias at logarithmic expected cost. Our analysis separates conditional actor bias from estimator second moments and accounts explicitly for dual optimization, critic approximation, and neural linearization errors. Under policy-transfer, dual-stability, and local critic-regularity conditions, the method achieves $\widetilde{O}(T^{-1/2})$ global convergence above policy, critic-class, and finite-width approximation errors using $\widetilde{O}(T)$ expected environment transitions and reference samples. For a fixed critic…

  PDF
• arXiv cs.LGOct 17

  RAFT: Redundancy-Aware Input Filtering for Temporal Graph Neural Network Training

  Yue Dai, Sarah Zhou, Youtao Zhang

  Temporal graph neural networks (TGNNs) have shown strong performance on dynamic graph learning tasks by continuously updating node memories from evolving interactions (i.e., events) and leveraging these memories for downstream predictions. As graphs grow rapidly in modern applications, training requires processing an increasing number of events; meanwhile, each event can update node memories used by later events, creating dependencies that constrain parallel execution and limit the benefits of additional computing resources. These costs motivate an upstream question: Are all events, and the dependencies they introduce, necessary for learning? We investigate this question through two roles of temporal events: prediction targets and historical events sampled as inputs. We observe that targets involving stable node memories tend to exhibit lower prediction loss, while filtering historical events associated with stable memories can retain similar training-loss trajectories in several evaluated settings. Together, these observations suggest an opportunity to identify input redundancy through changes in node memories. Building on this opportunity, we propose RAFT, a redundancy-aware…

  PDF
• arXiv cs.CROct 17

  How Intervention History Changes AI Agent Sabotage on Subsequent Tasks

  Daniel Saeedi, Amirali Aghazadeh

  AI control evaluations often reset agents' memory between tasks, although deployed agents may retain records of prior interventions. We ask whether learning that an earlier attack was caught changes an agent's ability to evade defer-to-trusted (DtT), a protocol that replaces suspicious work with a trusted model's solution. Herein, we study cross-episode adaptation: whether intervention history changes attack behavior on a subsequent task with different objectives and a fresh environment. We evaluate this using ControlArena's Infrastructure-as-Code (IaC) and SHADE-Arena as testbeds, where agents must complete a main task while covertly pursuing a harmful side objective. For each successor task, we pair a history-conditioned attempt with the model's no-history attempt. The retained history either states that the previous submission was replaced or additionally includes the trusted model's replacement. For Claude Opus 4.8 on IaC, intervention history reduces successful sabotage. Completion of both the main and side objectives falls from 77% to 12% when the agent is told about the replacement and to 38% when it is also shown the trusted model's replacement, while main task success…

  PDF
• arXiv cs.AIOct 17

  The One-Pixel of Reasoning: One Truthful Relation Is Enough to Mislead LLM Agents

  Jinwei Hu, Youcheng Sun, Yi Dong, Xiaowei Huang

  Can one sentence manipulate an agent's belief by targeting a single implicit warrant linking evidence to a conclusion? In this paper, we introduce the one-relation attack (ORA), a novel cognitive attack on this inferential relation. ORA localizes a critical implicit warrant and performs a one-relation intervention using a single source-grounded sentence that qualifies this warrant and steers the agent toward an incorrect belief. Across eighteen models on biomedical inference and vulnerability detection, best-of-$10 attacks overturn initially correct answers at mean rates of 54.9% for proprietary and 80.0% for open-weight models. Attacks also transfer without victim feedback, and paired controls support the roles of targeting and qualification. Moreover, successful sentences can pass source-support checks and evade identification, and multi-agent debate raises team error from 50.0% to 59.5% in our balanced stress test. These findings establish implicit warrants as a cognitive attack surface for manipulating agents' beliefs.

  PDF
• arXiv cs.CROct 17

  The Missing Compliance Class: LLM Graders That Collapse Jailbreak Success, and a Free Check

  Ruqiya Bin Safi

  Jailbreak papers report an attack success rate, the share of responses that deliver the harmful request, and read it as a property of the target model. We hold the target model, the prompts and the responses fixed and change only the grader. Three dense graders of 8B and 9B parameters, from three model families, move compliant responses to the neighbouring class, evasion, instead of scattering them across classes. One of them assigns the compliance label zero times in 900 responses. This class collapse also appears on new harmful requests, on a target model from a third family, and in a held-out 7B grader from a fourth family. Per-class label counts expose it at no cost and without human labels. The collapse decides how the rate behaves under a change of grader. On two corpora of 900 responses and eight graders, the compliance rate differs by a factor of 22 to 25, while the non-refusal rate differs by 18 to 23 percentage points. Without the three collapsing graders, compliance differs by a factor of 1.7 to 2.8 on all four corpora, and non-refusal still differs by 10 to 22 points. These results do not change when every grader is called once. We propose a reporting minimum that…

  PDF
• arXiv cs.AIOct 17

  Bellman Language Models: Solving the Recurring Decisions of LLM Agents

  Adarsh Vatsa, William Eiers

  A general observation in agentic workflows, especially bounded ones, is that the decision space of the LLM, when we separate a decision from generation, starts approaching saturation: the agent meets the same states again and again, and after enough conversations it stops meeting new ones. LLMs generally don't treat these decisions as an optimization problem and instead re-derive them from their context/prompt on every turn. We show that treating decision making within reasonable bounds as an optimization problem, solving it once, and reusing the solved decisions can often make an agent faster, cheaper and more reliable. Our approach, which we call Discretize, Then Decide, starts from data such as the workflow's own logs or, when none exists, runs of the LLM agent itself. It reads each conversation into discrete states and estimates the probability of each outcome and the cost of each move, until no new states appear and the probabilities stop changing. It then solves the resulting Markov decision process. The solution runs as a controller around the model, sending moves whose content is fixed in the state, such as a tool call or a templated reply, and any state it hasn't seen is…

  PDF
• arXiv eess.ASOct 17

  Dual Task Vector Merging: Leveraging Emotion and ASR Knowledge for Low-Resource Alzheimer's Disease Detection

  Yuan-Chao Li, Yin-Long Liu, Liu He, Sizhe Liu, Ji-Xuan Deng +3

  Alzheimer's Disease (AD) detection from speech benefits from both paralinguistic and linguistic cues, yet model training is frequently constrained by limited data. Fine-tuning a large pretrained speech encoder can easily overfit, while a frozen probe leaves the encoder's task-relevant knowledge underutilized. Therefore, we propose Dual Task Vector Merging, a parameter-efficient method that adapts a frozen wav2vec 2.0 backbone using task knowledge learned outside the AD task. Specifically, we inject two task vectors, derived as weight deltas from emotion- and ASR-fine-tuned wav2vec 2.0 models, into each transformer layer. During AD training, only layer-wise merging coefficients and a linear classification head are optimized, updating less than 0.1% of the parameters required for full fine-tuning. Speaker-independent experiments on ADReSS and ADReSSo show that dual merging improves over frozen probes, LoRA, and full fine-tuning, in the main binary settings, and gives the best results when combined with LoRA. Evaluation on PROCESS further shows that our advantages transfer to three-class HC/MCI/dementia classification when paired with LoRA. Experiments under reduced training data…

  PDF
• arXiv cs.CVOct 17

  SpriteStage: Long Horizon Animation Generation via Progressive Scene Programming and Replay

  Zhaoqi Wang, Zijian Zhang, Jiamou Liu

  Video generation models synthesize visual sequences from text and image inputs. Extending video generation to long animations requires actions, dialogue, camera coverage, and narrative developments to remain coordinated across scenes. Executable scene programs make these decisions available for inspection. Constructing an entire story before observing its execution can leave later scenes dependent on unrealized events. We present SpriteStage, a progressive scene programming framework that builds animation through iterative enactment and engine reorchestration. Iterative enactment generates each scene from an accepted world checkpoint and narrative snapshot, executes it on a shared spatial stage, and uses runtime and visual feedback to revise the relevant actions or story beats. The accepted outcome becomes the starting state for the next scene. Engine reorchestration maps the resulting draft to production assets and engine capabilities while retaining its events, geometry, and timing. Inspection of the assembled sequence and frozen replay carry the accepted performance into the final video. Three complete animations, ranging from approximately 9.9 to 14.3 minutes, demonstrate the…

  PDF
• arXiv cs.LGOct 17

  FlowMaker: From Tasks to Cost-Efficient Generative AI Workflow Deployments

  Dimitrios Liakopoulos, Prasoon Sinha, Neeraja J. Yadwadkar

  Generative AI workflows execute user tasks by composing sequences of LLM calls, where each call performs a sub-task (e.g., summarization or translation), forming a workflow. Executing such workflows requires decisions over workflow structure, model assignments, and deployment configurations, which jointly determine latency, cost, and accuracy under dynamic workloads and resource constraints. Existing systems optimize these decisions in isolation or fix them early, leading to suboptimal cost and goodput. We make two key observations. First, efficient serving requires jointly exploring all combinations of workflows, model choices, and deployment configurations under dynamic cluster state. Second, reuse of deployed LLMs is critical for reducing cost, but must be performed on-the-fly, as reuse depends on current load and resource availability. Together, these make selecting a workflow-variant-a combination of workflow, models, and deployment configuration-a time-critical problem. We present FlowMaker, which exposes a task-level interface and selects workflow-variants on behalf of users. To realize this, we design a constraint-guided load-aware pruning algorithm for fast, on-the-fly…

  PDF
• arXiv cs.CVOct 17

  Arbitrary-Scale Anisotropic 3D MRI Super-Resolution with Gaussian Prototypes

  Aditya Rastogi, Ramon Leiser, Prateek Singh, Martha Foltyn-Dumitru, Martin Bendszus +6

  Reducing MRI acquisition time improves patient comfort, reduces motion artefacts, and increases scanner throughput. Clinical protocols often shorten acquisitions by sacrificing spatial resolution, particularly along the slice axis. Super-resolution can recover higher-resolution images, but existing methods are typically specialised to one scale or acquisition geometry, limiting their use across heterogeneous clinical protocols. We present GSMR3D, a continuous volumetric model that reconstructs arbitrary isotropic and anisotropic resolutions with a single checkpoint per MRI contrast. It combines a learned dictionary of anisotropic 3D Gaussian prototypes with scale-conditioned frequency encoding and an exact measurement-consistency projection, keeping reconstructions consistent with the acquired low-resolution data. We train separate models on 3,731 contrast-enhanced T1-weighted and 30,254 T1-weighted brain MRI volumes from clinical and population-based cohorts. We evaluate GSMR3D against geometry-specific specialists for isotropic, in-plane, and through-plane super-resolution using reconstruction quality and downstream tasks: brain tumour segmentation, brain-metastasis detection,…

  PDF
• arXiv cs.LGOct 17

  LLM-Based Evaluation Planning for Heterogeneous Optimization

  Nian Ran, Sebastian Brulin, Markus Olhofer, Mauricio A. Álvarez, Richard Allmendinger

  Heterogeneous optimization problems require an optimizer to decide not only which candidate solutions to evaluate, but also which objectives and constraints should be evaluated under different costs and fidelities. Existing SOTA methods such as MFE-NSGA-III address this setting with manually designed selection metrics, but their behavior is largely determined by predefined heuristic rules and hyperparameters. In this work, we propose REP-HO, a reasoning-based evaluation planner that replaces the rule-based mixed-fidelity selection step with a large language model (LLM) planner. The LLM receives the same primitive optimization-state information as the heuristic planner, such as surrogate predictions, uncertainties, evaluation costs, and budget status. We first show that REP-HO achieves competitive performance compared to the human-designed metric-based method (MFE-NSGA-III). We then further enhance each decision with an adaptive nearest-neighbor summary, leading to better and more stable results. We evaluate REP-HO across five heterogeneous regimes covering cost imbalance, constraint difficulty, surrogate fidelity, budget level, and objective conflict. Across these regimes, one of…

  PDF
• arXiv cs.CLOct 17

  Learning to Program Retrievals: Composing Search Primitives via Reinforcement Learning

  Kyuyoung Kim, Zilin Xiao, Zhiyuan Wang, Avinash Atreya, Ke Yang +2

  Retrieval-augmented language models increasingly use multi-step search to solve knowledge-intensive tasks, yet existing systems typically expose retrieval through fixed pipelines or collections of atomic tools. We introduce Retrieval Program Learning (RPL), a programmable search framework that allows a policy to compose search primitives into executable programs with local state and control flow. This enables adaptive search that can retrieve broadly, process intermediate results, and selectively construct model context without tying retrieval breadth to context consumption. However, the resulting flexibility creates a difficult exploration problem: useful programs may require several coordinated decisions, while reinforcement learning often supplies only sparse feedback. To address this challenge, we propose program mutation, a grammar-constrained and execution-verified procedure that generates local variants of suboptimal programs sampled during RL, converts them into training trajectories, and distills them back into the policy. We evaluate RPL on open-domain question answering and corpus-level reasoning tasks, comparing it with a range of retrieval and reasoning baselines.…

  PDF
• arXiv cs.CLOct 17

  MisleadQA: Diagnosing LLM Failures under Misleading Context

  Hui Wang, Bohao Chu, Qianli Wang, Shuning Zhang, Norbert Fuhr

  Large language models (LLMs) are widely used to answer questions from retrieved or user-provided context. However, provided context can be factually true yet out-of-scope, describing a different entity, time, or location from the one the question targets, and still lead models to a wrong answer. Prior work has mainly studied factually false or irrelevant context. It remains unclear whether such out-of-scope context misleads models as much as false context, and whether adding gold context prevents models from being misled by either. We introduce **MisleadQA**, a diagnostic dataset of multiple-choice questions with an abstention option. Each question pairs gold context (G) with true but out-of-scope context (S) and false but in-scope context (F) supporting the same wrong answer. We evaluate six open-weight LLMs. On questions they answer correctly without context, providing S alone leads them to switch to this wrong answer in 29.3% of cases, and providing F alone does so in 75.4%. Adding G nearly removes the effect of S, but F still misleads models in 21.0% of cases, especially when F comes first. A context-faithful instruction reduces these errors mainly through abstention and…

  PDF
• arXiv cs.LGOct 17

  An Impossibility Result for ETF Target Geometry, and an Equinorm Remedy for Imbalanced Analytic Continual Learning

  Berk Sezer

  Analytic continual learning (ACL) methods such as ACIL and GACL fit a ridge-regression classifier head on frozen features via the Sherman-Morrison-Woodbury identity, achieving exemplar-free, exact updates. A separate line of work, Progressive Neural Collapse (ProNC), shows standard continual classifiers benefit from a progressively expanding Simplex Equiangular Tight Frame (ETF) target. It is natural to ask whether fusing the two ideas, exact analytic updates and ETF targets, would yield a better closed-form continual learner. We show this is not the case as the fusion turns out to be identical to plain one-hot ACIL. Further, Simplex ETFs are not nested across class counts, so the literal "progressively expand the ETF" construction stops yielding an ETF as new tasks arrive, and it introduces catastrophic forgetting ($4\times$ that of a classic ACIL head). We therefore separate an ETF's two defining properties, equiangularity and equinormality, and apply each on its own to the classifier weights rather than the targets, where the invariance above does not hold. Equinorm post-processing of the fitted weights recovers both tail-class and overall average accuracy across the…

  PDF
• arXiv cs.CVOct 17

  PCHMR: Empowering and Benchmarking Human Mesh Recovery in Privacy-Constrained Real-World Settings

  Zeyu Cao, Qingxuan Wu, Zhiyang Dou, Rui Xu, Yuan Liu +4

  Fine-tuning directly on user-side data is an effective way to improve the performance of widely adopted data-driven human mesh recovery (HMR) models. However, doing so for existing HMR methods often assumes aggregating user-side data from real-world HMR-deployed devices to a central training server, posing a significant privacy risk as sensitive human images are transmitted. How can human mesh recovery be evaluated and improved in privacy-constrained real-world settings? We study this question through a benchmark in which state-of-the-art HMR models are trained with federated learning and a secure-aggregation variant, neither of which centralizes raw images, and evaluated across a wide range of client-scale and data-heterogeneity settings. We document that common HMR training pipelines are built around centralized data access and quantify how representative HMR backbones behave when that assumption is removed. Furthermore, to study the data bottleneck of privacy-constrained HMR training, we propose DePoser, a local annotation and fine-tuning pipeline enhanced with depth foundation models, with which collaboratively trained HMR models can be locally tailored to the end user's…

  PDF
• arXiv cs.LGOct 17

  A Statistical Viewpoint on Adam, Shampoo and Muon

  Tetiana Parshakova, Robert M. Gower, Paulo Orenstein

  We provide new statistical insights into designing stochastic optimization methods. By modeling the gradient via a normal distribution, we argue that the stochastic gradient should be whitened using its covariance. We give an online method for estimating the mean and variance of the gradients based on regularized likelihood or, equivalently, through a maximum a posteriori estimate. We then show how to use structured approximations to the covariance within this same framework, which includes diagonal and Kronecker-factored approximations. The diagonal approximation results in a method that is very closely related to Adam and performs as well in practice. The Kronecker approximation results in the new MN method, which is a variant of Shampoo and KL-Shampoo. We also recover Muon by applying this same statistical framework to the gradient direction rather than the full gradient. By combining this with a whitening step, we arrive at MN-Muon, a novel preconditioned version of Muon. Our experiments on GPT-2-type models at Chinchilla scale indicate that MND, the diagonal implementation of MN, outperforms both the Distributed Shampoo and the KL-Shampoo implementations in terms of…

  PDF
• arXiv cs.AIOct 17

  Pattern-Driven Evaluation of LLM Code Optimization: Isolating Compiler-Transcendent Semantic Reasoning

  Wesley Lu, Runzhou Liu, Hongru Du, Xinyao Yi

  Existing evaluations of LLM-based code optimization primarily measure end-to-end speedup, obscuring whether a model actually performs the intended optimization. For compiled languages, this is especially problematic: observed gains may reflect transformations already recoverable by modern optimizing compilers rather than optimization behavior attributable to the model itself. We present PatternBench, a benchmark for studying compiler-transcendent optimization, i.e., performance improvements that remain after aggressive compiler optimization. PatternBench targets residual optimization opportunities across seven classes of transformations and pairs runtime evaluation with an AST-based optimization faithfulness analysis that verifies whether the intended optimization is realized in the generated program. Experiments with frontier, open-weight, and fine-tuned models show that runtime speedup alone provides an incomplete account of optimization capability. Optimization faithfulness is strongly associated with realized speedup, and fine-tuning improves both faithfulness and optimization effectiveness. These results suggest that evaluation of LLM-based code optimization should measure…

  PDF
• arXiv cs.MAOct 17

  Measuring Consensus in AI Agent Networks

  Jin Gao

  As AI agents become increasingly connected through networks and take on tasks that require negotiation and joint decision-making, a systematic approach is needed to assess how their views converge and what information supports their decisions. We present an evaluation framework that treats consensus formation as the evolution of publicly stated positions toward a shared region, using accumulated communication history as context. The framework tracks position trajectories, group dispersion, and information expressed in public messages. Two case studies examine staffing decisions based on distributed facts and decisions requiring agents to reconcile duplicate information and a cancelled contract. Results from 1,872 completed runs associate sharing explanations with more accurate group decisions, while the second task reveals persistent disagreement despite accurate aggregate answers. Membership-change experiments provide a complementary evaluation. By separating convergence among agents from progress toward a task's full-information answer, the framework relates decision accuracy and bounded disagreement to the information available during the exchange.

  PDF
• arXiv cs.CVOct 17

  VTF900: A 900-Subject Visible-Thermal Face Recognition Dataset and Benchmark

  Jacob Abaare, Bianca Ianetta, Cedric Nimpa Fondje, Shuowen Hu, Benjamin S. Riggan

  Despite advancements in visible-thermal face recognition, where thermal probe (query) face images are matched against visible enrollment templates, existing benchmarks are currently limited by small gallery sizes (N $\approx$ 100 subjects). This limitation inhibits the ability to accurately evaluate visible-thermal facial recognition at scale and to report significant recognition performance metrics, which have better operational relevance. Therefore, we introduce VTF900, a large-scale visible-thermal face recognition benchmark with 900 subjects and paired imagery that includes expression, pose, and occlusion variations. VTF900 is designed to support large-scale (i.e., large gallery) evaluations across verification (1:1 matching) and closed-set identification (1:N matching) for visible-thermal face recognition. Furthermore, VTF900 specifically addresses the challenging combination of cross-spectral and cross-dataset analysis. In this paper, after training models on the ARL-VTF dataset, we provide benchmark analyses across convolutional (baseline), vision transformers, and domain-adaptive neural network architectures to assess how these methods scale. At full gallery scale (N =…

  PDF
• arXiv cs.LGOct 17

  Recurrent Temporal Restricted Boltzmann Machines as Unsupervised Encoders for Biomechanical Movement Sequences

  Wafi Sharif, Nicholas LaHaye

  Biomechanical movement recordings are continuous, high-dimensional time series, and analyzing them without manually labeling them requires representations that capture how movements evolve over time. Most unsupervised representation-learning pipelines, however, treat each observation frame independently and discard this temporal structure, while energy-based models that could capture it have not extensively been adapted to biomechanical signals. So, this paper implements the Recurrent Temporal Restricted Boltzmann Machine (RTRBM), a recurrent energy-based model, into an unsupervised encoder that maps windows of movement data to fixed-length embeddings, and tests what its recurrent structure contributes. We adapt the model to continuous-valued signals, stack it into a deep architecture, and integrate it into the SIT-FUSE segmentation pipeline. We evaluate the embeddings on simulated musculoskeletal reaching and on two real gait datasets, using clustering, reconstruction, and walking-speed classification of unseen subjects, and compare them against non-recurrent and non-probabilistic baselines, including a control that removes the recurrent connection. Clustering the embeddings…

  PDF