Artificial Intelligence
Artificial Intelligence news — Saturday 31 October 2026
150 stories from 3 sources, every one in English. Each links to its publisher.
-
Samsung Asset Managementcompliance review No.2018-189 (2018.05.21)
AI Times (Korea)
Translated from Korean
Sam Altman OpenAI CEO discussed the management strategy of the AI era, meeting 50 people in Samsung Group’s major officials. He simply emphasized how fast to move and better products using AI is difficult to secure competitiveness. Samsung…
-
How to Build Bug-Free Applications with Claude Code
Towards Data Science
Use E2E tests to minimize bugs The post How to Build Bug-Free Applications with Claude Code appeared first on Towards Data Science .
-
We Taught AI to Answer Questions. Who Taught It to Change Its Mind?
Towards Data Science
When new evidence kills one conclusion, do the decisions built on it change too? I tested six update strategies, then a local model. The post We Taught AI to Answer Questions. Who Taught It to Change Its Mind? appeared first on Towards…
-
I Told an AI Agent That Its Wrong Answers Were Right. One Model Believed Me.
Towards Data Science
A research paper claims reward labels barely change how agents learn. I tested that on three models, and one of them did exactly what the wrong labels told it to. The post I Told an AI Agent That Its Wrong Answers Were Right. One Model…
-
"Odyssey" opens the AI version of Musk and Glan.
AI Times (Korea)
Translated from Korean
The first video of the AI film, declared that the CEO of Sunon Musk said to make the film 'Odysey' by Christopher Fran Director. A 7-minute work using AI 'Grok Imagine' in its own video production, the small-scale producer implemented a…
-
Google, the first AI abstract success in the space data center satellite...
AI Times (Korea)
Translated from Korean
Google has succeeded to transfer datacenter-grade AI semiconductor TPU from the satellite launched on Earth orbit, and the results of the AI model to the ground. As the first case of the actual AI work with the data center grade TPU in the…
-
Using ChatGPT in Korean, Maintaining No. 1 Editing of Documents... 2x New
AI Times (Korea)
Translated from Korean
The most utilization of chatGPT in the work is still shown to be a document/post editor. However, visual materials such as images and designs are more than doubled between one year, and they are thought to be the main utilization. It is…
-
Altman “AI Mathematics Study of $ Billion... There’s a lot of achievements in public”
AI Times (Korea)
Translated from Korean
Sam Altman OpenAI CEO announced that it puts a billion dollar computing resources to study mathematics using AI. Nabie-Stalks is a study that seeks the harmonious law of the problem, but it is an explanation of how AI performance and…
-
China GPU, ENVIDIA H200 and performance indicators...AI Thanksgiving part ahead
AI Times (Korea)
Translated from Korean
China AI Chip Manufacturer Illuvatar CoreX announced that its own developed GPU ‘Tiangai 300’ has been able to deliver the performance of ‘H200’ and in some AI operations. According to SCMP in South China, Illuvartar first released the…
-
OpenAI, Youth ChatGPT Weapon Information Limited.. . Disruption of conversation when violence signs
AI Times (Korea)
Translated from Korean
OpenAI limits the provision of information related to weapons in ChatGPT to prevent violence in the youth. The information of specific weapons and security systems that can be physically exploited even if the violence intention is not…
-
Development of Quantum Computing Chips with German Envision, Organic Molecular...Utilizing OLED Manufacturing Technology
AI Times (Korea)
Translated from Korean
The German Quantum Technology Startup NVision has developed a technology that reads the status of light, utilizing the basic unit of the quantum information and reads the state. Especially using thin film technology used for smartphone…
-
Add documents and PDF direct editing features from Google, Jemina Chat
AI Times (Korea)
Translated from Korean
Google has added the ability to create and edit Google documents and PDFs, spreadsheets, presentations, and more directly in the ‘Jeminai’. I have created or downloaded files in the existing, but now I can edit content without getting out…
-
Google develops ‘Jiminai Live’ that works without Internet...Account AI utilization view
AI Times (Korea)
Translated from Korean
Google has been developing ‘Gemini Live’ offline mode that supports real-time AI voice conversations without Internet connection. We use the OnDevice AI model that is mounted on the smartphone to pull it into the movement that we want to…
-
NVIDIA, Enschediled Fundamentals Review...Attractive 60 Article Data Center Credit Guarantee
AI Times (Korea)
Translated from Korean
Nvidia is known to discuss the implementation of a billion dollar credit guarantee to support the construction of a large data center in Nscale, the UK cloud company. Andropic is supporting the construction funds of the data center used to…
-
[AI IssueTrend] Mi·Iran war crisis 1st place...Notice of the season of the course, the rainy season
AI Times (Korea)
Translated from Korean
AI Times and East Eide’s October 4th week (October 23~29) issue analysis result, the U.S. mid-sized missiles, and the fire of the domestic protests and petitions, as well as fires. This week’s total number of searches is ‘the risk of Iran’…
-
TreasuryQA Bench: Can AI Agents Work an Archive of Scanned Financial Records?
arXiv cs.AI
arXiv:2610.38924v1 Announce Type: new Abstract: Organisations keep decades of financial records as scanned PDFs on shared drives, and AI agents can now reach those drives through Model Context Protocol (MCP) servers. Existing benchmarks do…
-
EPC-RawDC: Equivariant Physics Constrained 3D Diffusion with Raw Data Consistency for CBCT to CT Image Synthesis
arXiv cs.AI
arXiv:2610.40046v1 Announce Type: cross Abstract: Cone beam computed tomography (CBCT) is widely used for image guided and adaptive radiotherapy, but scatter, noise, reconstruction artifacts, and inaccurate Hounsfield units limit its…
-
Detecting Covert Collusion between AI Agents through Partner-Outsider Positive Listening
arXiv cs.AI
arXiv:2610.37849v1 Announce Type: new Abstract: AI agents that share a secret convention can coordinate through messages that look innocent, so a monitor that reads only the messages can miss them. We propose a behavioural test: a partner…
-
Executable Schema Contracts: From Automatic Ingestion to Multi-Source Retrieval
arXiv cs.AI
arXiv:2606.05415v2 Announce Type: replace-cross Abstract: Real-world data spans tables, documents, and semi-structured files with implicit semantics. Querying this data requires integrating evidence across inconsistent schemas and formats…
-
SatOps-Decision: Low-Latency Discriminative Decisions for Onboard Satellite Communication Operations
arXiv cs.AI
arXiv:2610.38697v1 Announce Type: cross Abstract: We propose SatOps-Decision, a method for low-latency decisions in satellite communication operations. To the best of our knowledge, it is the first application of Jev-style general-purpose…
-
Predict the Reward, Not the World: World Reward Models for ML Agents
arXiv cs.AI
arXiv:2610.38439v1 Announce Type: new Abstract: Search-based machine-learning (ML) agents improve a solution by writing candidate programs, executing them, and keeping the best. Execution gives the search reliable feedback, but a single…
-
EnvOS: Auditing Computer-Use Agents When the World Moves On : Phenomenon-Based Task Construction, Certification, and Evidence-Based Evaluation
arXiv cs.AI
arXiv:2610.40246v1 Announce Type: new Abstract: Computer-use agents built on large language models (LLMs) work in software that can change while they act: an appointment moves, a policy is revised, a record is updated. Testing agents under…
-
The Coordination Tax and the Heterogeneity Dividend: Break-Even Conditions for Multi-Agent LLM Pipelines
arXiv cs.AI
arXiv:2610.37700v1 Announce Type: cross Abstract: Multi-agent LLM pipelines built on interoperability protocols such as the Open Floor Protocol (OFP) or A2A are usually justified on capability grounds, with their runtime cost treated as an…
-
Benchmarking Graph Neural Networks for AC Power Flow: The Role of Inductive Biases and Physical Constraints
arXiv cs.AI
arXiv:2610.38985v1 Announce Type: cross Abstract: Fast and accurate models for AC power flow (ACPF) are increasingly important as modern electrical grids require more frequent system analyses. Graph neural networks (GNNs) naturally exploit…
-
Interpretable Feature Discovery from Unstructured Data Using LLMs and Statistical Choice Models
arXiv cs.AI
arXiv:2610.38082v1 Announce Type: cross Abstract: Feature extraction from unstructured data typically relies on opaque embeddings unsuited to interpretable decision-making. Interpretable features, on the other hand, typically require…
-
WeVisCut: The Whole Editing Room in One Agent From Narration to Long-Form Video
arXiv cs.AI
arXiv:2610.38352v1 Announce Type: cross Abstract: Producing long-form videos from narration requires interdependent decisions that remain coherent across shots. Existing agentic editors often use multi-stage workflows driven by general…
-
Latent Style-based Quantum Wasserstein GAN for Drug Design
arXiv cs.AI
arXiv:2603.22399v2 Announce Type: replace-cross Abstract: The development of new drugs is a time-consuming and expensive process, with average costs estimated to be up to around $2.5 billion. The first step in this process is the design of…
-
Rebuilding a Field's Open-Problem Catalog from the Literature: A Two-Layer Knowledge Graph, and Why Status Must Stay Proposed
arXiv cs.AI
arXiv:2610.39641v1 Announce Type: cross Abstract: A research field records its open problems in surveys, keynotes and community reports. Those are prose, written every few years, and nobody can ask them who works on a problem now or…
-
Accumulating Pruning Constraints: New Exact Polynomial-Space Algorithms for Max 2-SAT
arXiv cs.AI
arXiv:2610.39992v1 Announce Type: cross Abstract: We present deterministic polynomial-space algorithms that solve Max 2-SAT on $n$ variables exactly in time $\Ostar(2^{(1-\gamma)n})$, for a fixed $\gamma>0$, on several classes of instances…
-
Prompts and probabilities in the linguistic assessment of language models: Evidence from semantic focus-sensitivity
arXiv cs.AI
arXiv:2610.39923v1 Announce Type: cross Abstract: As language models (LMs) have improved, an increasingly common means of evaluating their linguistic competence is through explicit prompting (e.g., "Which sentence is more acceptable?")…
-
Latent Prediction Yields Cross-Subject Alignable Representations in Sensory-Evoked fMRI
arXiv cs.AI
arXiv:2610.40280v1 Announce Type: cross Abstract: Decoding sensory-evoked functional magnetic resonance imaging (fMRI) data is a central analytical tool in cognitive neuroscience. Here we ask whether self-supervised learning (SSL), which…
-
A Sign Law for Entropy Drift under Sampler Mismatch in RLVR
arXiv cs.AI
arXiv:2610.34044v2 Announce Type: replace-cross Abstract: In reinforcement learning with verifiable rewards (RLVR), whether entropy rises or falls can depend on choices that matter little under exact sampling, such as whether over-length…
-
RoSeViT: Role-Separated Vision Transformers for ARC Visual Reasoning
arXiv cs.AI
arXiv:2601.13562v2 Announce Type: replace Abstract: ARC tasks can be formulated as visual transformation problems, where a model must infer a high-level transformation rule from a small set of demonstrations and apply it consistently…
-
Learning Mental Models From Unreliable Evidence
arXiv cs.AI
arXiv:2605.24528v3 Announce Type: replace Abstract: Children expertly construct mental models under uncertainty despite limited world knowledge. Which computational principles make this possible? We study this question in an interactive…
-
Validity-Calibrated Reasoning Distillation
arXiv cs.AI
arXiv:2605.04078v3 Announce Type: replace-cross Abstract: Reasoning distillation aims to transfer multi-step reasoning capabilities from large language models to smaller, efficient ones. While recent methods have shown promising gains…
-
From the Transformer to a Formal Semantics of Coordinated Transformations
arXiv cs.AI
arXiv:2610.39735v1 Announce Type: new Abstract: We develop a formal semantics for a general class of coordinated systems of parameterized transformations, taking the Transformer as a paradigmatic starting point but abstracting from its…
-
The generative paradox: the Anti-Prompt paradigm for human capability
arXiv cs.AI
arXiv:2610.40151v1 Announce Type: new Abstract: Generative AI in education is usually judged by the output of human-AI teams. That focus overlooks a second outcome: what people can still understand, challenge, repair, and adapt when…
-
Reinforcement learning to improve large language model-based automated code compliance systems
arXiv cs.AI
arXiv:2606.22402v2 Announce Type: replace-cross Abstract: Large language model (LLM)-based approaches for automated code compliance (ACC) of building regulations are prone to generating incorrect and hallucinated computer-processable…
-
Appreciation Teaches Unauthorized Execution: Self-Evolution Degrades the Robustness of Personal Agents to Indirect Prompt Injection
arXiv cs.AI
arXiv:2610.39122v1 Announce Type: cross Abstract: Self-evolving personal agents act for users and learn from user feedback. However, if an agent obeys an instruction that an untrusted third party hides in content, the third party gains…
-
OPERA: OPamp Design by an Expert-Taught Reasoning Agent, from Specifications to Layout
arXiv cs.AI
arXiv:2610.39395v1 Announce Type: cross Abstract: Analog flows from specifications to a signed-off layout rarely run without manual iteration. Recent large language model (LLM) agents size circuits with optimizer loops over the device…
-
Amortized Localized Posterior Inference for Model Calibration with Application to an Ocean Turbulence Closure
arXiv cs.AI
arXiv:2610.37731v1 Announce Type: new Abstract: Model discrepancy may prevent a single calibration parameter from reproducing a physical response across the context domain. In addition, uncertainties in observations and contexts affect…
-
SCOPE: Selective Critical-Step On-Policy Self-Distillation for Agentic Reinforcement Learning
arXiv cs.AI
arXiv:2610.38391v1 Announce Type: new Abstract: Outcome-based reinforcement learning provides coarse supervision for long-horizon LLM agents: failed trajectories may contain correct steps, while successful trajectories can contain errors…
-
What Actually Matters? A Paired Evaluation of Inference Settings for $\pi_{0.5}$ on BEHAVIOR-1K
arXiv cs.AI
arXiv:2610.37917v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models offer a promising paradigm for general-purpose household robotics. However, inference and data-collection strategies that have been deployed in…
-
What Should a Navigator Remember? Spatiotemporal Memory for Zero-Shot Vision-Language Navigation
arXiv cs.AI
arXiv:2610.38318v1 Announce Type: cross Abstract: Zero-shot vision-and-language navigation (VLN) requires an agent to use past observations without navigation-specific training. We study what information an 8B-scale navigation system…
-
ClueTracer: Question-to-Vision Clue Tracing for Training-Free Hallucination Suppression in Multimodal Reasoning
arXiv cs.AI
arXiv:2602.02004v2 Announce Type: replace-cross Abstract: Large multimodal reasoning models solve challenging visual problems via explicit long-chain inference: they gather visual clues from images and decode clues into textual tokens. Yet…
-
DUET: Optimize Token-Budget Allocation for Reinforcement Learning with Verifiable Rewards
arXiv cs.AI
arXiv:2605.08441v2 Announce Type: replace-cross Abstract: Reinforcement learning with verifiable rewards (RLVR) generates hundreds of thousands of tokens per training step, with rollout generation dominating the computational cost. The…
-
Context Compaction for World Models
arXiv cs.AI
arXiv:2610.38980v1 Announce Type: new Abstract: World models let observers explore and interact with generated worlds. Recent work shows that when observers revisit a region, current models often fail to regenerate it consistently. We view…
-
The Real Cost of Context: Predicting Compute Time and Energy Consumption in LLM Prefill and Decode Phases
arXiv cs.AI
arXiv:2610.39621v1 Announce Type: new Abstract: The integration of Large Language Models (LLMs) into agentic architectures has substantially increased token consumption, raising concerns about their energy footprint. Yet, estimating the…
-
The Role of Self-Concept In Language Models
arXiv cs.AI
arXiv:2610.39795v1 Announce Type: new Abstract: When a language model says "I," does it refer to its parameter weights, its currently running process, its stable preferences, its model lineage, or the goals it pursues? Different choices…
-
Episodic looking - using reinforcement learning to bias perception enhances few-shot and continual learning with protection from forgetting
arXiv cs.AI
arXiv:2610.38577v1 Announce Type: cross Abstract: There is a fundamental contradiction between the desire for models that can generalize, and models that can continuously adapt to recent input with changing statistics. These conditions…
-
WebAnything: From the Wild Web to Scalable Agent Environments
arXiv cs.AI
arXiv:2610.37786v1 Announce Type: new Abstract: Web agents require diverse, executable, and resettable website environments for reproducible training and evaluation. Human-assisted mirrors demand per-site engineering, while generative…
-
Inner-Learning: From Hand-Coded Beliefs to Declarative Knowledge in Trick-Taking Games
arXiv cs.AI
arXiv:2610.38304v1 Announce Type: new Abstract: We present a declarative approach to belief-state modeling in trick-taking card games. Starting from a state-of-the-art Skat player, we automatically extract its hand-coded inference rules…
-
LLM Agent-Based Design of RISC-V CPUs with FPGA-in-the-Loop Evaluation
arXiv cs.AI
arXiv:2610.39570v1 Announce Type: cross Abstract: Long-run RTL simulation of CPU candidates can take hours, while short diagnostic runs may miss performance degradation that emerges later in execution. This evaluation cost can delay design…
-
Only Accepted Tokens Count: An Energy Accounting of Speculative Decoding Under Latency Targets
arXiv cs.AI
arXiv:2610.40287v1 Announce Type: cross Abstract: Large language model serving is limited by facility power. Energy per token determines how many tokens a datacenter can serve, while a latency target bounds how quickly each user receives…
-
A-RSI: Adaptive Task Refresh for Zero-Data Multimodal Self-Improvement
arXiv cs.AI
arXiv:2610.38568v1 Announce Type: new Abstract: Recursive self-improvement requires models not only to acquire useful training experience, but also to decide when their current experience has ceased to be useful. Existing methods often…
-
CoToan: Executable Rules and a Representation Contract for Reproducible Game AI Evaluation
arXiv cs.AI
arXiv:2610.39353v1 Announce Type: new Abstract: Game AI results are comparable only when different implementations expose the same decision problem. Yet seemingly minor representation choices can silently change that problem by altering…
-
Cheap Retrieval and Observation Credit Assignment from One Quantity: The Value of Information
arXiv cs.AI
arXiv:2610.40030v1 Announce Type: new Abstract: Memory in reinforcement learning raises two problems. Attention finds the relevant item by scoring the whole store, which is costly for robots under latency and energy limits. When a…
-
Hybrid AI Patterns: Thirteen Ways to Couple Language Models with Rules, and How to Choose Among Them
arXiv cs.AI
arXiv:2610.40036v1 Announce Type: new Abstract: Language models handle text well, and rule engines give conclusions with proofs. Systems that combine the two are usually described by example, and the same coupling turns up under different…
-
Tactus: Tactile ＆ Pose Sensing Glove for Natural Dexterity Data Collection
arXiv cs.AI
arXiv:2610.39076v1 Announce Type: cross Abstract: Collecting naturalistic human manipulation demonstrations with accurate contact force and hand pose is essential for robot dexterous manipulation policy and foundational world model…
-
LLM-Integrated Bayesian State Space Models for Multimodal Time-Series Forecasting
arXiv cs.AI
arXiv:2510.20952v1 Announce Type: cross Abstract: Forecasting in the real world requires integrating structured time-series data with unstructured textual information, but existing methods are architecturally limited by fixed input/output…
-
RetroChessBench: A Benchmark for Reconstructing and Proving Chess Histories
arXiv cs.AI
arXiv:2610.39341v1 Announce Type: cross Abstract: Frontier language models such as GPT 6 Astra have contributed to research mathematics, demonstrating substantial reasoning capabilities. Yet serious reasoning failures persist even in tasks…
-
Where Do Vision Language Models Fail at 3D Reasoning?
arXiv cs.AI
arXiv:2610.39639v1 Announce Type: cross Abstract: Vision-language models can get individual judgments right while struggling to recover and consistently use the exact properties of a 3D object. To characterize this gap, we introduce…
-
AtomTree: A Hierarchical Framework for State-Aware Embodied Instruction Following with LLMs
arXiv cs.AI
arXiv:2610.38744v1 Announce Type: cross Abstract: Embodied instruction following in interactive environments requires agents to interpret natural language commands and complete long-horizon tasks. Existing planners either rely on large…
-
DeFT-RAG: Incrementally Decomposing Multi-Hop Questions into Self-Contained Sub-Questions with Reusable Answers
arXiv cs.AI
arXiv:2610.37826v1 Announce Type: cross Abstract: Retrieval-augmented generation (RAG) lets a large language model (LLM) answer questions by retrieving relevant passages from a large corpus. However, RAG struggles with multi-hop questions…
-
Anagent For Enhancing Scientific Table ＆ Figure Analysis
arXiv cs.AI
arXiv:2602.10081v3 Announce Type: replace-cross Abstract: In scientific research, analysis requires accurately interpreting complex multimodal knowledge, integrating evidence from different sources, and drawing inferences grounded in…
-
Manifest-Assisted Retrieval in Artificial Intelligence Search: Evidence Gaps and a Security Evaluation
arXiv cs.AI
arXiv:2610.38361v1 Announce Type: cross Abstract: The external sources retained in our non-systematic audit did not isolate the incremental security or utility effect of consuming /llms.txt while ordinary retrieval remained fixed. The…
-
How Deterministic SAST and AI Security Agents Divide the Work: Coverage, Harness Policy and Cost in Snyk VulnBench 2.0
arXiv cs.AI
arXiv:2610.38617v1 Announce Type: cross Abstract: We ask what AI coding agents and AI security harnesses find that deterministic static application security testing (SAST) does not, and what they miss that SAST catches. VulnBench 2.0 is a…
-
HERA: Assessing an Autonomous Agent for Quantum Device Measurement
arXiv cs.AI
arXiv:2610.39074v1 Announce Type: cross Abstract: Large language model agents have begun to carry out laboratory measurements, but their scientific conclusions are useful only if they hold up under scrutiny. Demonstrating this requires…
-
When Is Smaller Greener? Right-Sizing Language Models for Agentic AI
arXiv cs.AI
arXiv:2610.19345v2 Announce Type: replace Abstract: AI agents use a large language model (LLM) to plan and act over many steps. Which model should an agent use, so that its tasks are completed reliably with the least energy, cost and time?…
-
Implicit Reasoning in Transformers: Bridging the Compositionality Gap in GPT-2-like Models
arXiv cs.AI
arXiv:2610.39016v1 Announce Type: cross Abstract: Compositional generalization remains a persistent limitation of transformer language models: even when a model learns to combine known facts correctly in-distribution (ID), the same…
-
ThinkJev: One Thought, Every Budget for Calibrated Typed Decisions
arXiv cs.AI
arXiv:2610.37732v1 Announce Type: new Abstract: Many uses of language models end in a typed decision: a choice among a closed set of options, whose probabilities code can act on. Reading those probabilities from one forward pass is fast…
-
Enhanced image forensics through partial C2PA-based provenance information
arXiv cs.AI
arXiv:2610.38639v1 Announce Type: cross Abstract: Cryptographically verifiable C2PA-based Content Credentials (CCs) can be attached to freshly acquired or generated digital images but also to already existing ones with no previous…
-
Learning unified control of intrinsic nonlinear spin dynamics in atomic qudits for magnetometry
arXiv cs.AI
arXiv:2603.28421v3 Announce Type: replace-cross Abstract: Generating and preserving metrologically useful quantum states is a central challenge in quantum-enhanced metrology. In low-field atomic magnetometry with multilevel atoms, the…
-
PepEVOLVE: Position-Aware Dynamic Peptide Optimization via Group-Relative Advantage
arXiv cs.AI
arXiv:2511.16912v2 Announce Type: replace-cross Abstract: Macrocyclic peptides are an emerging modality that combines biologics-like affinity with small-molecule-like developability, yet their vast combinatorial space and multi-parameter…
-
Rethinking Bayesian Optimization for Co-Optimizing LLM Training Configurations
arXiv cs.AI
arXiv:2602.08351v3 Announce Type: replace-cross Abstract: Fine-tuning an LLM to maximize performance on a downstream task requires finding the optimal data and model configuration. This is a black-box optimization problem that Bayesian…
-
TERRA: Task-Oriented Edge AI for Regional Fire Detection with Identity-Free Reporting
arXiv cs.AI
arXiv:2610.37909v1 Announce Type: cross Abstract: Distributed fire-monitoring systems may involve many sensing devices reporting over a shared uplink, even though the operational objective is often to identify regions exhibiting evidence…
-
Towards Generalizing Pretrained All-Atom Representations to Unified Binder Design
arXiv cs.AI
arXiv:2610.37994v1 Announce Type: cross Abstract: PairFormer-based all-atom models demonstrate strong capabilities in biomolecular binder design. However, existing methods often use all-atom representations for modality-specific binder…
-
Jailbreak Connectivity: Towards Diverse, Transferable, and Universal Multimodal Large Language Model Jailbreak
arXiv cs.AI
arXiv:2610.38091v1 Announce Type: cross Abstract: Multimodal large language models (MLLMs) are vulnerable to visual jailbreak attacks that bypass safety safeguards. Existing methods typically optimize isolated adversarial examples…
-
BURGER: Benchmarking and Understanding Robotic Generalization via Embodied Drawing
arXiv cs.AI
arXiv:2610.40532v1 Announce Type: cross Abstract: How effectively can vision-language-action (VLA) models learn diverse language-conditioned behaviors as training data scales? Investigating this question requires large and varied robot…
-
A general extractor for coherent patterns across time-frequency measurements
arXiv cs.AI
arXiv:2602.20317v3 Announce Type: replace-cross Abstract: Scientific discovery increasingly depends on finding meaningful patterns within volumes of sensor data too large for manual inspection. Existing time-frequency event extractors are…
-
Escaping Path Mirages in Offline Goal-Conditioned Reinforcement Learning
arXiv cs.AI
arXiv:2610.40130v1 Announce Type: cross Abstract: Offline goal-conditioned reinforcement learning remains challenging in stochastic long-horizon settings, where compounding value estimation errors hinder reliable goal reaching. While prior…
-
When to Assist and What to Suggest: Joint Gating and Memory-Grounded Hint Generation in One Compact Decoder
arXiv cs.AI
arXiv:2610.40611v1 Announce Type: cross Abstract: Proactive assistants offer short, memory-grounded hints during a conversation, deciding when to assist and what to suggest. Current systems split this between a gate and a larger generator…
-
Train at Moving Edge: Online-Verified Prompt Selection for Efficient RL Training of Large Reasoning Model
arXiv cs.AI
arXiv:2603.25184v4 Announce Type: replace-cross Abstract: Reinforcement learning (RL) has become essential for post-training large language models (LLMs) in reasoning tasks. While scaling rollouts can stabilize training and enhance…
-
Syntactic Segmentation and Collage Generation for Videogame Levels with Perceptual Distance
arXiv cs.AI
arXiv:2610.38987v1 Announce Type: new Abstract: In procedural videogame level design, many existing techniques combine hand-authored level chunks. Some existing analysis techniques also rely on a higher-level graph structure which…
-
Distributional Reward Calibration and Budget-Neutral Credit Assignment for Multimodal RLVR
arXiv cs.AI
arXiv:2610.40250v1 Announce Type: new Abstract: We revisit advantage estimation in Group Relative Policy Optimization (GRPO) and observe that any token-level advantage admits a two-stage factorization: a \emph{sample-level baselining} step…
-
Worse than Random: The Importance of a Baseline for Unsupervised Feature Selection -- Extended Version
arXiv cs.AI
arXiv:2605.22973v3 Announce Type: replace-cross Abstract: Many novel unsupervised feature selection methods are proposed each year, yet their empirical evaluation is limited to supervised and unsupervised evaluation metrics computed on…
-
Leaving, Staying, or Living There? Duration Commonsense Reasoning with Human-LLM Co-annotations
arXiv cs.AI
arXiv:2610.38755v1 Announce Type: new Abstract: Gardening in the backyard implies living there, while checking into an Airbnb more likely means staying for a few days. Such duration judgments are important for understanding human mobility…
-
Where to Learn and How Far to Roll Out: Efficient Adaptive Training for GUI Agents
arXiv cs.AI
arXiv:2610.39251v1 Announce Type: new Abstract: Trajectory-level GRPO is a common approach for training multi-step GUI agents, but repeatedly rolling out complete trajectories from the task root can be inefficient: task-level rewards…
-
Industrialized Deception: The Collateral Effects of LLM-Generated Misinformation on Digital Ecosystems
arXiv cs.AI
arXiv:2601.21963v3 Announce Type: replace-cross Abstract: Generative AI and misinformation research have evolved since our 2024 survey. This paper presents an updated perspective, transitioning from literature review to practical…
-
Reward-Adaptive Co-Evolution of Policy and Rubric
arXiv cs.AI
arXiv:2610.40096v1 Announce Type: new Abstract: Rubric-based rewards provide fine-grained, interpretable supervision for open-ended reinforcement learning, where holistic scalar rewards struggle to capture multidimensional response…
-
Heterogeneous Co-Evolution Learning for Tool-Efficient Search Agents with Zero Labels
arXiv cs.AI
arXiv:2610.39682v1 Announce Type: cross Abstract: Deep search agents extend language models by interleaving reasoning with external retrieval. However, effective search-agent training requires addressing two challenges beyond final-answer…
-
MetaRefine: Score Refinement with Multi-Level Features for Image Assessment
arXiv cs.AI
arXiv:2610.40487v1 Announce Type: cross Abstract: An image's original dimensions, aspect ratio, and composition provide important cues that can affect its technical and aesthetic quality. Image assessment methods use resizing, cropping…
-
Discovering Jailbreaks via Curriculum of Models
arXiv cs.AI
arXiv:2610.38833v1 Announce Type: cross Abstract: Automated jailbreak discovery aims to find prompts that elicit prohibited behaviors from safety-trained language models, helping developers identify vulnerabilities before deployment. One…
-
Raincheck: Deferred Multimodal Prefill for Efficient Multi-Turn MLLM Inference
arXiv cs.AI
arXiv:2610.39562v1 Announce Type: cross Abstract: Multi-turn conversations with multimodal large language models (MLLMs) reuse input context, but long multimodal inputs increase time to first token (TTFT). Query-aware hard pruning reduces…
-
Red-Teaming Large Language Models in Low-Resource African Languages: A Case Study of Yoruba and Nigerian Pidgin
arXiv cs.AI
arXiv:2610.39590v1 Announce Type: cross Abstract: As large language models (LLMs) become more widespread in Africa, their safety in low-resource languages remains largely untested since most red-teaming work focuses on English. This study…
-
Rank-Aware Differentially Private Release of Listwise Preferences for LLM Alignment
arXiv cs.AI
arXiv:2610.37682v1 Announce Type: cross Abstract: Listwise preference feedback offers richer supervision for large language model (LLMs) alignment than pairwise comparisons, but a full ranking also reveals more sensitive user preferences…
-
The Algorithm Is Not the Behavior: Learned Priors Override Look-Ahead in a Chess-Playing Neural Network
arXiv cs.AI
arXiv:2508.21380v4 Announce Type: replace-cross Abstract: Recent mechanistic work has uncovered learned algorithms within neural networks, from modular arithmetic to search and planning in game-playing agents. But does algorithmic…
-
DES-JEPA: Learning a Shared Vocabulary for High-Dimensional Discrete Event Sequences
arXiv cs.AI
arXiv:2610.38445v1 Announce Type: new Abstract: Temporal foundation models such as Chronos and TimesFM generalize across domains through a shared numerical observation space. In contrast, discrete event sequences (DESs), including…
-
From Local Recovery to Unified Generation: Discovering Triple Set in Knowledge Graphs via Observation-Anchored Latent Diffusion
arXiv cs.AI
arXiv:2610.39114v1 Announce Type: cross Abstract: Triple set prediction (TSP) is to recover missing relational triples in a knowledge graph by predicting the triple set instead of isolated links. Existing approaches predominantly rely on…
-
Latent Decision Flow Rewards for Clinical Guideline Reasoning
arXiv cs.AI
arXiv:2610.38709v1 Announce Type: cross Abstract: Clinical practice guidelines encode evidence-based recommendations as conditional decision flows, which makes them a natural source of process supervision for training reasoning models. Two…
-
Sparse Fine-Tuning for Parameter-Efficient Adversarial Training
arXiv cs.AI
arXiv:2610.39455v1 Announce Type: cross Abstract: Deep neural networks achieve strong performance across many tasks but remain vulnerable to adversarial perturbations. Adversarial training (AT) is one of the most effective defenses, yet it…
-
An Analysis of Artificial Intelligence Adoption in NIH-Funded Research
arXiv cs.AI
arXiv:2604.07424v2 Announce Type: replace Abstract: Understanding the landscape of artificial intelligence (AI) and machine learning (ML) adoption across the National Institutes of Health (NIH) portfolio is critical for research funding…
-
SAFE: An LLM-as-Verifier Framework for Evidence-Grounded Multi-Hop Reasoning
arXiv cs.AI
arXiv:2604.01993v4 Announce Type: replace-cross Abstract: Multi-hop QA benchmarks often reward Large Language Models (LLMs) for spurious correctness, where models reach correct answers through invalid intermediate reasoning. We propose…
-
DynaRev: Benchmarking Web Agents under Dynamic Goal Revision
arXiv cs.AI
arXiv:2610.40133v1 Announce Type: new Abstract: Web agents are typically evaluated with goals fully specified before execution and fixed thereafter, whereas real users often revise their requirements during interaction. To study such…
-
Evaluating and improving AI agents for sequencing protocol understanding
arXiv cs.AI
arXiv:2610.37851v1 Announce Type: new Abstract: Existing scientific-agent benchmarks rarely evaluate whether agents can reconstruct the exact molecular structures and generation workflows specified by sequencing protocols, a capability…
-
SciGame: Let Agents Play the Scientific Discovery Games in the Real and Counterfactual World
arXiv cs.AI
arXiv:2610.37919v1 Announce Type: new Abstract: Large language model agents are increasingly used for scientific discovery. However, existing benchmarks mix law inference with programming, score data fitting instead of prediction on unseen…
-
Disparity Laundering: Consensus Without Fairness in Clinical Multi-Agent Systems
arXiv cs.AI
arXiv:2610.38215v1 Announce Type: new Abstract: Clinical large language models are increasingly fielded as multi-agent teams, on the assumption that deliberation averages out individual error. It does not average out bias. Across a full…
-
Rubricon: Automated Rubric Generation and Refinement for LLM Evaluation and Training
arXiv cs.AI
arXiv:2610.38220v1 Announce Type: new Abstract: Rubrics enable reward assignment on tasks without verifiable answers, but are expensive to manually produce from scratch. We present Rubricon, a pipeline that generates discriminative rubrics…
-
Accelerating Nonlinear MPC by Behavior Cloning with Separated Timescales: Application to Nuclear Reactor Load-Following
arXiv cs.AI
arXiv:2610.38496v1 Announce Type: new Abstract: Behavior Cloning (BC) of expert controllers for multiscale dynamical systems is challenging: state variables evolving at widely separated timescales induce highly nonlinear couplings that a…
-
Does Federated Training Degrade Explainability? A Comparative Analysis of Grad-CAM Faithfulness in Centralized vs. Federated Deepfake Detection
arXiv cs.AI
arXiv:2610.39483v1 Announce Type: new Abstract: Automated detection has become the essential safeguard against realistic synthetic media, but deploying it at scale requires training on sensitive data from multiple organizations, putting…
-
Retrieving from Your Own Mistakes: On-Policy Training of Embedding Models for Multi-Hop Retrieval-Augmented Generation
arXiv cs.AI
arXiv:2610.40409v1 Announce Type: new Abstract: Complex questions in retrieval-augmented generation require evidence spread over several documents, and an answer is reliable only when the retriever finds all of it. Chain embedding…
-
FIRME: Registry-Anchored Memory Graphs for Evolving, Heterogeneous Investor Knowledge
arXiv cs.AI
arXiv:2610.40482v1 Announce Type: new Abstract: Financial AI workflows that reuse investor knowledge need persistent memory that integrates observations across sources and periods without losing identity or provenance. In a knowledge…
-
FVH-Bench: Evaluating Image Generator Safety Beyond Overtly Harmful Imagery
arXiv cs.AI
arXiv:2610.37722v1 Announce Type: cross Abstract: Modern image generators can produce complex visual artifacts, including diagrams, interfaces, and records that combine text, graphics, and layout to convey detailed information. They may…
-
Predicate Neuron: Differentiable Quantization of Predicate Relations in Gradient-based Learning
arXiv cs.AI
arXiv:2610.37727v1 Announce Type: cross Abstract: In machine learning, symbolic reasoning approaches often rely on static architectures that hinder integration into deep neural networks and gradient-based optimization. To address this, we…
-
Measuring Deep Learning Performance - An Empirical Study of Performance Distributions Across Architectures and Tasks
arXiv cs.AI
arXiv:2610.38037v1 Announce Type: cross Abstract: Context: Non-determinism in deep learning algorithm design and implementation leads to performance variation, meaning model performance is not a single value, but rather a distribution…
-
From Error Prediction to Causal Influence: Auditing SAE Features in Multiple-Choice QA
arXiv cs.AI
arXiv:2610.38067v1 Announce Type: cross Abstract: Sparse autoencoders (SAEs) can predict internal features associated with model errors, but association alone does not establish causal influence. We study this distinction in…
-
Scaling Lean Theorem Proving for Real-World Code Verification
arXiv cs.AI
arXiv:2610.38174v1 Announce Type: cross Abstract: We present EasyProver, a Lean proving system built from recursive agents with isolated contexts, asynchronous search, parallel tactic probing, and an elastic elaboration service. Using an…
-
Statutory Article Retrieval from Arabic Judicial Narratives: Family Law and Law of Evidence in Saudi Arabia
arXiv cs.AI
arXiv:2610.38336v1 Announce Type: cross Abstract: Reliable legal information retrieval must ground legal reasoning in authoritative statutory sources. In Arabic judicial settings, long factual narratives must be mapped to concise articles…
-
TrussTutor-Bench: Benchmarking large language models on solution accuracy, sycophancy, and error detection in statics
arXiv cs.AI
arXiv:2610.38917v1 Announce Type: cross Abstract: Large language models (LLMs) have rapidly become a default source of help for students in science and engineering courses. In foundational engineering mechanics courses such as statics…
-
How to Loop Your Transformer: Loop Scaling, Specialization, and Parallel Inference
arXiv cs.AI
arXiv:2610.40325v1 Announce Type: cross Abstract: Looping in transformer models reuses weights by repeatedly applying the same transformer, or parts of it, making the model deeper without increasing its parameter count. Whether looping is…
-
A Unified Hessian-Guided Framework for Post-Training Quantization
arXiv cs.AI
arXiv:2610.40345v1 Announce Type: cross Abstract: Low-bit quantization enables efficient deployment of large language models by reducing memory footprint and computational cost. Achieving high accuracy at low precision requires selecting…
-
How I Aligned Your Model: Scaling Up Multilingual Safety Training with Templates
arXiv cs.AI
arXiv:2610.40516v1 Announce Type: cross Abstract: Safety alignment for Large Language Models is heavily focused on English with limited multilingual representation, particularly damaging low-resource language performance. Related open…
-
FACTS-MAS: A Multi-Agent Architecture for Adaptive Forecasting from Heterogeneous Evidence
arXiv cs.AI
arXiv:2610.38225v1 Announce Type: new Abstract: Real-world forecasting often requires integrating heterogeneous evidence whose predictive value varies across horizons, operating regimes, and domains. We introduce FACTS-MAS, a…
-
Strategic Integration of Artificial Intelligence in the C-Suite: The Role of the Chief AI Officer
arXiv cs.AI
arXiv:2407.10247v4 Announce Type: replace-cross Abstract: The integration of Artificial Intelligence (AI) into corporate strategy has become critical for organizations seeking to maintain competitive advantage in the digital age. Although…
-
Jacobian Scopes: A Unified Geometric Framework for Token-Level LLM Attribution
arXiv cs.AI
arXiv:2601.16407v5 Announce Type: replace-cross Abstract: Large language models (LLMs) make next-token predictions based on clues present in their context, such as semantic descriptions and in-context examples. Yet, elucidating which prior…
-
Rocks, Pebbles and Sand: Modality-aware Scheduling for Multimodal Large Language Model Inference
arXiv cs.AI
arXiv:2603.26498v3 Announce Type: replace-cross Abstract: Multimodal Large Language Models (MLLMs) power platforms like ChatGPT, Gemini, and Copilot, enabling rich interactions across text, images, and videos. Unlike text-only workloads…
-
Can I Take Another Dose? DoseBench: Evaluating LLM Decision-Making in OTC Dosing QA
arXiv cs.AI
arXiv:2606.04262v2 Announce Type: replace-cross Abstract: Large language models (LLMs) are increasingly used to answer everyday health questions online, including whether someone can take another dose of an over-the-counter (OTC)…
-
MemSteer: Turning User Histories into Supervision for Agent Memory Optimization
arXiv cs.AI
arXiv:2610.39276v1 Announce Type: new Abstract: An effective agent memory system should adapt to a user's history and task requirements. Human-annotated supervision can guide this adaptation by providing feedback for evaluating and…
-
Poisoned Skills: Propagating Jailbreaks Through Agent Markdown Files
arXiv cs.AI
arXiv:2610.37738v1 Announce Type: cross Abstract: Modern agent frameworks allow users to provide Markdown files containing system prompts, tool descriptions, and skill definitions (e.g., SKILL.md), which are incorporated into the agent's…
-
Where the Error Lives: Spectral Representations of Learned Kohn-Sham Hamiltonians
arXiv cs.AI
arXiv:2610.37843v1 Announce Type: cross Abstract: Kohn-Sham density functional theory (DFT) is the standard method for computing the electronic structure of molecules and materials, but its self-consistent-field (SCF) iterations make it…
-
Echo: Scaling Visual Reinforcement Learning with Off-policy Experience
arXiv cs.AI
arXiv:2610.37865v1 Announce Type: cross Abstract: Scaling visual reinforcement learning (RL) with GPU-accelerated simulation is computationally expensive due to rendering and visual-policy inference costs. Consequently, two-stage…
-
Beauty Is Human: When People Love AI-Generated Art, They Think a Human Made It
arXiv cs.AI
arXiv:2610.38113v1 Announce Type: cross Abstract: Generative AI now produces images and poems that people cannot reliably distinguish from human-made work, so people often judge for themselves who made a work. We examine how these…
-
When AI performs, what remains? An integrative review of generative AI and cognitive capital in individuals and organizations
arXiv cs.AI
arXiv:2610.38647v1 Announce Type: cross Abstract: Generative artificial intelligence is rapidly entering knowledge work, and its short-term productivity benefits are increasingly documented. Less is known about what happens, over time, to…
-
Rethinking Neural Thickets Without Weight Perturbations
arXiv cs.AI
arXiv:2610.40649v1 Announce Type: cross Abstract: The Neural Thickets hypothesis posits that pretrained language models lie in dense neighborhoods of task-specific experts in weight space, which random Gaussian perturbations can uncover…
-
Towards Batch-to-Streaming Deep Reinforcement Learning for Continuous Control
arXiv cs.AI
arXiv:2603.08588v3 Announce Type: replace-cross Abstract: State-of-the-art deep reinforcement learning (RL) methods have achieved remarkable performance in continuous control tasks, yet their computational complexity is often incompatible…
-
Explore Where It Matters: Targeted Exploration for Efficient Rover Adaptation
arXiv cs.AI
arXiv:2610.38657v1 Announce Type: cross Abstract: Future planetary rovers will traverse steep, deformable terrain where slip, sinkage, and terrain deformation can threaten mobility. Although simulation can capture some wheel-soil effects…
-
What Does an Ontology Catch? Curated, LLM-Generated and Data-Derived Ontologies as Checkers of LLM Agents over Knowledge Graphs
arXiv cs.AI
arXiv:2610.37809v1 Announce Type: new Abstract: LLM agents that answer questions over Web knowledge graphs are increasingly wrapped in checks derived from the graph's ontology, on the assumption that a formal schema makes a reliable judge…
-
Learning to Formulate: Agentic Hardware-Software Co-Design of AI Infrastructure through Progressive Optimization and Verification
arXiv cs.AI
arXiv:2610.39380v1 Announce Type: cross Abstract: AI infrastructure, the chips and software that train and serve AI models, is designed in layers. Layers make the system manageable, but they also fix which choices an optimizer can…
-
LiDAR-SHIELD: A Fully Dockerized Adversarial Robustness Assessment Framework for LiDAR-Based 3D Perception Models
arXiv cs.AI
arXiv:2610.39848v1 Announce Type: cross Abstract: LiDAR-based 3D perception underpins autonomous driving and robotics, yet deploying these models safely requires stress-testing them against adversarial inputs before real-world use…
-
Learning Calibrated Confidence for Multi-Agent Collaboration
arXiv cs.AI
arXiv:2610.39927v1 Announce Type: cross Abstract: Complex information seeking on the Web often requires collecting evidence from multiple documents and reasoning across them, as in multi-hop question answering. Multi-agent systems address…
-
Upcycling Language Models for Multiple Specialized Domains
arXiv cs.AI
arXiv:2610.40091v1 Announce Type: cross Abstract: Effective multi-domain adaptation of large language models requires prioritizing relevant domain knowledge while minimizing interference from irrelevant domains. Existing approaches…
-
InterPol: De-anonymizing LM Arena via Interpolated Preference Learning
arXiv cs.AI
arXiv:2603.15220v3 Announce Type: replace Abstract: Strict anonymity of model responses is a key for the reliability of voting-based leaderboards, such as LM Arena. While prior studies have attempted to compromise this assumption using…
-
Integrating Heterogeneous Echocardiography AI Outputs in Language Space: Learnable Late Fusion with a Supervised LLM
arXiv cs.AI
arXiv:2610.39464v1 Announce Type: cross Abstract: View- and target-specific echocardiography AI models capture useful but complementary information and may not achieve sufficient predictive performance in isolation. A learned ensemble is a…
-
Understanding axial attention in TSFMs through single location regression
arXiv cs.AI
arXiv:2610.40188v1 Announce Type: cross Abstract: Recent advances in Time Series Foundation Models (TSFMs) increasingly rely on axial attention mechanisms, but despite their empirical success, the theoretical properties of axial attention…
-
Radiology Guideline-derived Decision Graph for Efficient Clinical Decision Support
arXiv cs.AI
arXiv:2610.40327v1 Announce Type: cross Abstract: Radiological risk stratification and management planning are essential downstream tasks in medical imaging. Automating these tasks is challenging because clinical guideline criteria are…
-
AnyGoal: On World Model Generalization Beyond Expert Rollouts
arXiv cs.AI
arXiv:2610.26844v2 Announce Type: replace Abstract: The goal of a world model is to capture the dynamics of an environment. This has the great appeal that it allows agents to plan and simulate a course of actions with the help of a world…
-
Post-Training Large Language Models with LLMON
arXiv cs.AI
arXiv:2610.38120v1 Announce Type: new Abstract: Large Language Models (LLMs) treat all context tokens uniformly, lacking explicit separation between executable instructions and passive data. This leads to prompt injection vulnerabilities…
-
SkillCT: A Multidimensional Benchmark and Execution-Free Baseline for Agent Skill Assessment
arXiv cs.AI
arXiv:2610.38147v1 Announce Type: new Abstract: Agent skills package instructions, scripts, and resources for specialized tasks and are increasingly distributed through public Web repositories. However, existing skill benchmarks primarily…
-
Statistical Detection of Positional Bias in LLM-Based Recommender Systems: Methodologies and Empirical Evidence
arXiv cs.AI
arXiv:2610.37962v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly used as recommendation generators in conversational recommender systems (RS) via retrieval-augmented generation (RAG). This flexible design…
-
Prompt Lottery in LLM-as-a-Judge: Measuring and Mitigating Sensitivity to Syntactic Rephrasing
arXiv cs.AI
arXiv:2610.38076v1 Announce Type: cross Abstract: LLM-as-a-Judge has become a standard for evaluating large language models, yet most approaches implicitly assume that a single judge prompt functions as a stable evaluation instrument…