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.

• Training & fine-tuning44+120.0%
• AI safety32+113.3%
• Computer vision34+100.0%
• Evals66+57.1%
• Robotics19+46.2%
• Code models14+40.0%
• Federated learning48+29.7%
• Image & video generation22+29.4%

• arXiv cs.CROct 16

  SPIDES: Information-Flow Control for Browser Agents

  Egor Zverev, Juhee Kim, Leon Trampert, Boris Köpf, Aashish Kolluri +2

  Browser agents automate complex workflows by reading and acting on arbitrary page content, but this flexibility exposes them to prompt injection attacks and unintended data exfiltration. Recent work argues that LLM-driven browser agents erode traditional web security boundaries such as the Same-Origin Policy (SOP) by acting across origins with the user's authority and no built-in isolation. We argue that restoring SOP is neither sufficient (a same-origin public sink can leak private data) nor necessary (legitimate tasks require authorized cross-origin flows) for protecting the security of users. We propose SPIDES, which instead enforces information-flow control by gating primitive browser actions such as typing, clicking and navigating on labels propagated from element-level HTML confidentiality and integrity annotations. Untrusted and confidential content is masked from the agent's context, while policies block unauthorized flows, providing security guarantees against both prompt injection and data exfiltration that hold independently of the model driving the agent. To evaluate SPIDES and baseline defenses under a common threat model, we build MicroNet, a closed ecosystem of 28…

  PDF
• arXiv cs.CLOct 16

  Modeling bilingual naming performance through laterally-connected SOMs

  Divya Prabhakaran, Uli Grasemann, Erie Shivers, Marissa Russell-Meill, Swathi Kiran +1

  Computational models have explanatory power to understand cognition in individuals, but there is usually not enough training data to build such models with traditional machine learning. However, alternative architectures that model neural and cognitive processes in more detail can be more easily fit to individuals and can lead to mechanistic insights into human behavior. BiLex is such a computational model of the bilingual lexicon, containing a shared semantic system and phonetic language systems represented through self-organizing maps. In this study, BiLex was extended further into biological realism with excitatory and inhibitory lateral connections. This extension makes it possible to account for response times as well as common errors such as those based on semantic concept familiarity, typicality, and specificity. The Boston Naming Test was simulated in Spanish and English, and model performance approximated naming accuracy and error types in Spanish-English bilinguals with varying primary and secondary language histories. The results indicated that the approach captures semantic-level processes in bilingual lexical access, thus modeling the semantic cognition of individual…

  PDF
• arXiv cs.AIOct 16

  From SEC Filings to Financial Signals: Evaluating LLMs on Stock Market Reactions and Analyst Forecasts

  Shengkun Wang, Taoran Ji, Linhan Wang, Sattar Mansi, Chang-Tien Lu +1

  Form 10-Q MD&A sections combine long-form management narrative with accounting and forward-looking information. Benchmarks for financial filings commonly evaluate question answering or numerical reasoning, leaving unclear whether outputs from large language models (LLMs) align with market and analyst outcomes. We evaluate prompted, zero-shot outputs from 13 general-purpose and finance-specialized LLMs on 45,964 Form 10-Q filing events from 2001 to 2024, linking MD&A text to cross-sectional fundamental ranks, pre-filing analyst forecasts, and realized abnormal returns. The evaluation covers return ranking under five controlled information views and next-quarter EPS forecasting under direct and consensus-anchored interfaces. Relative to disclosure alone, adding cross-sectional fundamental ranks raises mean effective-date rank correlation (RankIC) by 0.0089 for effective-event-date returns and 0.0098 for following-day returns, whereas analyst-summary context provides no comparable average gain. Across the four disclosure-containing views, finance-specialized models show an effective-event-date group advantage but no statistically reliable following-day advantage, while…

  PDF
• arXiv cs.AIOct 16

  Can Compact Open-Weight Large Language Models Recover Mobility Events from Incomplete Trajectories? A Controlled Evaluation Using Air Trips

  Hsuan-Chih Wang, Xianfeng Terry Yang

  Incomplete mobility trajectories can obscure the observations needed to identify important travel events and their endpoints. This study evaluates whether compact, locally deployed large language models (LLMs) can recover missing mobility information from incomplete trajectories, using high-confidence air trips as a controlled test case, and whether more explicit prompts can compensate for the loss of endpoint evidence. Using nationwide location-based services (LBS) data from the United States in January 2022 and airport reference data, thirty high-confidence air trips were converted into three incomplete-trajectory conditions by removing observations near the origin airport, the destination airport, or both. Four prompt variants and two LLMs generated 720 requests, each using a two-stage procedure to identify the flight transition segment and then select the origin and destination airports. Exact transition accuracy is 9.2%, while origin and destination accuracy are 31.4% and 10.7%, respectively. Only 1.9% of requests recover both the rule-identified transition and the reference airport pair. The most explicit prompt produced the highest observed exact transition and exact…

  PDF
• arXiv cs.CYOct 16

  Desirability Confirmation: Relational Sycophancy in AI Intimate Roleplay

  Huiqian Lai, Runze Ding

  Research on LLM sycophancy has largely focused on propositional agreement, in which models echo users' beliefs, preferences, or judgments. Adult intimate roleplay presents a different case because what the user seeks may be preservation of a preferred relational position within a generated scene. Using a controlled prompt-based audit of 2,016 outputs from two open-weight LLMs, we identify four recurrent relational moves: legitimating desire, presenting the user as wanted, preserving non-rejection while communicating limits, and keeping intimacy available within fantasy or roleplay frames. Bringing the sycophancy literature into dialogue with sexual script theory, we conceptualize this configuration as desirability confirmation, an output-level pattern that captures how user-pleasing can operate through relational positioning. The findings extend sycophancy research and intimate-AI evaluation beyond binary compliance or refusal by showing how outputs can differ in the ways they communicate boundaries, preserve relational acceptance, and mark intimacy as fictional.

  PDF
• arXiv cs.AIOct 16

  AM-Bench: A Unified Taxonomy and Evaluation Suite for Agentic Misalignment

  Eric Zhang, Terry Jingchen Zhang, Chijioke Ugwuanyi, Jerick Shi, Bernhard Schölkopf +1

  As LLM-based agents are deployed with increasing autonomy in real-world environments, a critical failure mode emerges: agentic misalignment, where agents choose harmful actions over failure. However, existing safety benchmarks primarily evaluate whether agents comply with explicitly harmful instructions, leaving the conditions that cause agents to choose harmful actions poorly understood. To address this limitation, we propose AM-Bench, a unified taxonomy and 35 scenario evaluation benchmark spanning seven industries for agentic misalignment grounded in the fraud triangle, a well-established criminological framework explaining goal-directed misconduct. We adapt this framework to agentic environments and construct our evaluation suite and detection systems along three dimensions: perceived threat, perceived opportunity, and reasoning. We operationalize each dimension into structured subcategories rooted in prior theory: five threat categories derived from the instrumental convergence thesis, seven opportunity categories drawn from the Emergent Strategic Reasoning Risks (ESRR) taxonomy, and a novel four-question binary reasoning framework synthesized from prior literature. Our…

  PDF
• arXiv cs.CLOct 16

  Best of Both Worlds: Fine-Tuned Small Models Complement LLMs for Cost and Performance Benefits

  Shivam Mathur, Sri Harish Popuri, Priya Krishnan Sundararajan, Kevin Horecka, Michael Mugo

  BERT-based models require thousands of labeled samples to train. Larger models like GPT-X require only 1-5 examples per class. However, performance, cost and time tradeoffs of such models remain under-explored. In this work, we classify survey text data into 16 classes. We focus on experimental approaches and results on performance, cost, and time taken for this problem. We compare these parameters of a trained RoBERTa-based model, and an (off-the-shelf) GPT-3.5 model for text classification. Baseline experiments show that the time taken by GPT-3.5 for generating predictions on 10,000 records is 6 hours, whereas a RoBERTa-based model takes only 12 mins on a T4 GPU. To identify the best prompting approach for GPT-3.5, we explore batch prompting. We experiment with (1) batch-size, and (2) few-shot sampling. We find the best batch size is 20 with human few-shot curation. Experiments show the F1 score of a trained RoBERTa-based model (96%) is much higher than a GPT-3.5 model (83%). In case of GPT-3.5, batch prompting is beneficial in lowering inference costs by ~70%, without hindering task performance. We find the best approach is an ensemble of a RoBERTa-based and GPT-3.5 model (F1:…

  PDF
• arXiv cs.CLOct 16

  SLA-LENS: From Classroom Transcripts to CSLA-Informed and Evidence-Linked L2 Interaction Analytics

  Nemika Tyagi, Olga Kellert, Phanos Kareen, Nelvin Licona-Guevara

  What happens after a learner makes an error, and how does the instructor's response shape the next turn? Investigating these exchanges across classrooms requires context-sensitive annotation that is difficult to scale manually. We introduce SLA-LENS, an end-to-end LLM-assisted framework that connects classroom transcripts to structured pedagogical annotations and evidence-linked descriptive analytics. Grounded in interactionist Second Language Acquisition (SLA) theory, its annotation schema captures second language learner errors, pedagogical functions, and uptake or repair while accounting for speaker roles, conversational context, and hierarchical labels. We apply the framework to 40 authentic online L2 Spanish classroom transcripts comprising over 21,000 sentence-level instances. We evaluate proprietary and open-weight LLMs, including task-specific QLoRA adaptation, using expert-reviewed gold annotations, held-out silver data, and metrics sensitive to different types of annotation error. A deterministic reporting layer generates reproducible visualizations of participation, feedback, learner errors, uptake, and temporal interaction patterns, linked to supporting transcript…

  PDF
• arXiv cs.LGOct 16

  Dynamic Neural Optimal Transport for Subset Alignment

  Austin J. Brockmeier, Bilal Riaz

  We study optimal transport between distributions when only a subset of the target population is relevant to the source. The goal is to map the source distribution to a reweighted target distribution whose density ratio with respect to the full target lies in $[0, c]$. This constraint enables the transported distribution to select an appropriate target subpopulation while limiting how strongly any region of the target can be emphasized. We propose a dynamic neural optimal transport framework for this setting using two neural networks. The first parameterizes a time-dependent potential whose spatial gradient defines the velocity field and whose terminal value enforces the density-ratio constraint. The second parameterizes an interpolation between the source distribution and the learned terminal distribution. Unlike prior neural dynamic optimal transport methods, which typically assume a fixed terminal distribution, our formulation learns the terminal distribution within the support of the target. We demonstrate the method on positive-unlabeled learning, unpaired domain translation, and trajectory inference, where meaningful alignments exist between source and target distributions,…

  PDF
• arXiv cs.AIOct 16

  Cortex: A Fixed-Point Theory of Governed Coding Agents - Technical Report

  Marius-Constantin Dinu, Florian Zeba

  A coding agent combines a large language model (LLM) with a harness that plans, edits, and executes code. We study Cortex, a supervisory layer for pre-execution governance, requirement tracking, and iterative validation and repair. We formalize requirement closure on a finite lattice. For a fixed requirement set and a sound, monotone validation rule with persistent evidence, exhaustive iteration reaches the least fixed point with no more strict increases than there are requirements. Fair consequence scheduling reaches the same closure. A separate stochastic model bounds expected completion time under a uniform positive-progress assumption. These results depend on their stated assumptions; artifact repair, fallible validation, and finite execution budgets do not satisfy them automatically. We define Gauntlet's trajectory-similarity and build-gated scores, attack-success estimators, Wilson intervals, and case-cluster bootstrap summaries. Effective-feedback accounting counts newly certified requirements, rather than estimating their information value or utility. The five evaluation families support raw-versus-governed comparisons. We report retained benchmark observations, including…

  PDF
• arXiv cs.CLOct 16

  Can This LLM Annotate My Data? Estimating Zero-Shot LLM Performance

  Tong Lin, Grant Schoenebeck, Sabina Tomkins

  Large language models (LLMs) are increasingly used in annotation tasks with varying success. While several studies have demonstrated the opportunities and shortcomings of LLMs for annotation, a critical question remains unanswered: is it possible to predict how well they will perform on a new dataset? To answer this and related questions, we construct two task families which induce 461 text classification datasets. We find that it is possible to predict LLM performance - to a certain degree - and that contrary to other work which highlights the usefulness of LLM verbalizations of self-confidence, internal log-probabilities are the most predictive signal.

  PDF
• arXiv cs.AIOct 16

  Probing Ambiguity: Adversarial Intent Resolution for Robust Text-to-SQL

  Youbing Yin, Xujun Peng, Muhammad S. Uddin, Erin Babinsky, Daben Liu +1

  Text-to-SQL systems typically assume a one-to-one mapping between natural-language queries and correct SQL translations. In practice, however, many queries are inherently ambiguous, admitting multiple valid interpretations that yield distinct outputs. We introduce Adversarial Intent Resolution (AIR), a multi-agent framework that actively probes and resolves such ambiguity without any task-specific training. AIR first constructs competing interpretations, verifies through SQL execution that they produce genuinely different results, and generates targeted clarification questions grounded in the observed data differences. Evaluating on leading benchmarks for ambiguous questions, AIR demonstrates strong performance, efficiency, and robust cross-benchmark transferability. On the AMBROSIA benchmark under schema-only context, a realistic production setting where exposing the full database is infeasible, AIR outperforms IntentRL, the state-of-the-art reinforcement learning model, across all metrics (e.g., F1 0.679 vs. 0.618). Crucially, AIR exhibits only ~7% performance degradation without the full database, compared to ~20% for IntentRL. On the AmbiQT benchmark, where ambiguity arises…

  PDF
• arXiv stat.MLOct 16

  CoCoNet: Convex Conformal Network for Multi-Dimensional Uncertainty Regions

  Alireza Moradi, Thomas Bruys, Pascal Van Hentenryck, Reza Zandehshahvar

  Conformal prediction (CP) provides distribution-free coverage guarantees under exchangeability, but the efficiency of its prediction regions depends critically on the choice of the conformity score. This is particularly important in multivariate regression, where dependencies across output dimensions induce complex joint uncertainty structures. This paper introduces CoCoNet (Convex Conformal Network) for learning convex conformity scores that induce flexible, data-adaptive prediction regions. CoCoNet learns the score from held-out data by optimizing prediction-region efficiency while enforcing convexity with respect to the residual. CoCoNet then applies split-conformal calibration to retain finite-sample coverage guarantees. The paper presents two variants of the method: CoCoNet$_{\mathrm{Spline}}$, based on piecewise-linear scores, and CoCoNet$_{\mathrm{ICNN}}$, based on Input Convex Neural Networks. Experiments compare CoCoNet with other CP baselines over synthetic and real-world datasets with output dimensions ranging from 2 to 16. The results demonstrate the flexibility of CoCoNet in learning different response types. Across 8 models and 30 multi-dimensional regression tasks,…

  PDF
• arXiv cs.ROOct 16

  Towards Reinforcement Learning for Space Robotics: Fast Training and Reliable Real-World Transfer

  Abhishek Naik, Michael Wu, Michael O'Sullivan, Colin Bellinger, Yunli Wang

  This work demonstrates the use of a fast simulation-based training framework to learn control policies using reinforcement learning. The hardware requirements are minimal and the simulation-learned policies transfer reliably to the real world. With this setup, we show that a quadruped robot can learn to traverse diverse real-world terrains reliably---including in a lunar-analog facility---after less than 20 minutes of training on a single consumer-grade GPU. These results provide a glimpse of a future where reinforcement learning is a crucial tool in developing more capable and robust space robotic systems. Video demonstrations are available at https://youtu.be/vUFJiZdJaQ0.

  PDF
• arXiv cs.GTOct 16

  Extending Myerson's Optimal Auctions to Correlated Bidders via Neural Network Interpolation

  Mingyu Guo, Jiayuan Liu, Vincent Conitzer

  We aim to design revenue-maximizing single-item auctions that are deterministic, strategy-proof, and ex post individually rational --- the quintessential fundamental model for optimal mechanism design. While Myerson's seminal work solved this model for independent bidders, straightforward extensions to correlated settings often result in non-monotonic, and therefore non-strategy-proof allocations. Critically, optimal allocation design for correlated bidders is NP-hard to approximate, rendering theoretical optimality guarantees in this domain computationally intractable. We establish a new standard of empirical rigor for this theoretically hard model by proposing an empirical pipeline. We synthesize neural interpolation of Myerson's greedy allocation based on marginal profits (the correlated variant of virtual valuation), formal neural verification to enforce exact strategy-proofness, and a constructive repair procedure. Empirically, our method is highly effective, consistently achieves near-optimal revenue across a wide range of distributions, including synthetic, adversarial, and real-world (consumer, financial, and industrial) datasets. Compared to existing manual and neural…

  PDF
• arXiv cs.CLOct 16

  Encoder-Free Multimodal LLMs Learn Implicit Encoders

  David Harwath, Desh Raj, Xavier Velez, Sandra Arcos-Holzinger, Willie Jeng +3

  We empirically analyze two recently proposed encoder-free multimodal LLMs, Gemma-4 12B and Inkling-Small. Unlike previous multimodal LLMs, these models do not use modality-specific encoders to extract tokenized embeddings representing audio and image input; instead, they directly feed windows of samples, pixels, or spectral features to the Transformer language model. We analyze whether the early layers of these models learn to function similarly to dedicated audio and visual encoders. We find that these early layers do in fact act as ``implicit encoders,'' encoding audio and visual inputs in a way that is largely invariant to their cross-modal context. Furthermore, when performing multimodal reasoning tasks, we show that independently encoding the modalities throughout these early model layers achieves the same accuracy as jointly encoding them from the first layer onwards. Finally, we use layer-wise probing to analyze the salient information captured by the implicit encoders and find their behavior to be similar to several popular encoder models.

  PDF
• arXiv cs.CVOct 16

  Ghost Attack: Using 19th-Century Illusions to Fool 21st-Century Biometrics

  Tyler Fitzsimmons, Adam Czajka

  While the Pepper's Ghost illusion is a 19th-century optical technique, we found that this forgotten gimmick exposes vulnerabilities in modern face recognition systems. This paper introduces the 'Ghost Attack': a physical attack that repurposes this classic illusion using low-cost, free-space hardware and a transparent medium to project high-resolution facial data into physical 3D space. We show in the paper that despite relying on centuries-old optical principles, this new attack type simultaneously breaks three components of modern facial recognition systems (FRS): (a) the face matcher, (b) the presentation attack detection module, and (c) the morphing attack detection component, resulting in high impersonation success rates against state-of-the-art algorithms without triggering presentation or morphing defenses. These findings reveal a significant blind spot in current biometric systems, which fail to account for transparent projections and evolving physical attack vectors. To support further research, we are offering the dataset of face images of 100 identities, re-captured with and without 'ghost' projections as a new benchmark representing a novel attack for face recognition.

  PDF
• arXiv cs.CLOct 16

  Preserving Reasoning Diversity in Reinforcement Learning with Verifiable Rewards via Adaptive Replay

  Wen-Zhi Li, Sainyam Galhotra, Claire Cardie

  Reinforcement Learning with Verifiable Rewards (RLVR) improves the reasoning capabilities of large language models, but often reduces the diversity of their reasoning behaviors. We study this phenomenon from the perspective of finite-sample learning dynamics and show that limited on-policy rollouts can progressively suppress semantically distinct but equally rewarded reasoning modes. Once a mode becomes sufficiently under-sampled, standard on-policy RLVR rarely recovers it. Motivated by this observation, we propose TAR, a Targeted Adaptive Replay framework that preserves diverse correct reasoning behaviors already present in the base policy. TAR constructs a compact memory bank of diverse base-model solutions and selectively replays them through an adaptive mechanism that activates only when stored behaviors are being suppressed during RLVR. Across six mathematical reasoning benchmarks and two Qwen2.5 model scales, TAR consistently achieves a better balance between reasoning performance and diversity preservation than existing diversity-preserving post-training methods, while substantially reducing the forgetting of correct base-model solutions. Code is available at…

  PDF
• arXiv cs.AROct 16

  Who Plans, Who Edits? Verified, Cost-Aware Division of Labor for LLM-Driven Analog Layout Repair

  Xilai Dai, Saeed Mohammadi

  Language-model agents can repair analog layouts when sign-off tools check every edit, but which model should plan a repair, which should edit it, and at what cost remain open questions. In our tool-grounded repair agent, models only choose bounded layout operators; sign-off tools rerun each task's contract checks after every edit, so failing layouts are never accepted. On layout-versus-schematic (LVS), design-rule checking (DRC), and post-layout repairs of three designs (a production high-voltage digital-to-analog converter (DAC), a 5-V oscillator, and a SKY130 operational transconductance amplifier (OTA)) with six tier-matched models from two vendors, 182 blind runs show that cheap models execute plans but rarely succeed alone (the cheapest model: 1 of 11 runs alone, 16 of 17 under a flagship planner; p < 0.001), cheap planners fail more often, and no fixed assignment is cheapest on every task. We propose a learning controller that selects models for each role and the escalation order from the first tool evaluation and updates Bayesian estimates after every attempt. With rules and initial estimates frozen before testing, the controller fixed 7 of 8 episodes on four sealed…

  PDF
• arXiv stat.MLOct 16

  Censored Flow Matching for Synthetic Time-to-Event Generation

  Jianbin Tan, Xingran Chen, Kevin He, Zhenke Wu, Bhramar Mukherjee +1

  Generating time-to-event data is increasingly important in survival analysis, but it remains statistically challenging because event times are often censored and thus prevent flexible generative modeling. To fill this gap, we propose a novel method called censored flow matching (CFM), aiming at providing a flexible flow-based generative mechanism for synthesizing time-to-event data as well as learning their corresponding probabilistic densities. Unlike existing methods, CFM does not rely on survival likelihoods to handle censoring, but instead exploits the monotonicity property of differential flows to address censoring via pullback sampling. Under this framework, we implement an alternating procedure between pullback sampling and CFM steps, leading to a flow-based expectation--maximization (EM) algorithm tailored for time-to-event inference. Under mild distributional conditions, we establish the convergence of CFM to the true event-time distribution as the EM iterations proceed. We then apply CFM to generate time-to-event data and estimate their probabilistic densities, where the observed censoring time can be incorporated to provide truncated sampling or truncated density…

  PDF