AX·NEWS
Sign in

Artificial Intelligence

Artificial Intelligence news — Friday 20 November 2026

150 stories from 7 sources, every one in English. Each links to its publisher.

Open in the live feedRSS feedOn a TV

  1. Antropic mathematics "Proof of the 167th anniversary of the clroad"... Notice
    KRAI Times (Korea) Translated from Korean

    Andropic’s mathematics claimed to prove the ‘Riemann Hypothesis’, one of the largest mathematics in the mathematics that were not full for 167 years. If you’re certified by the mathematics, you’ll be a case that AI will create a historical…

  2. [Nov. 20] Google "AGI is not the end"...AGI will present four development routes
    KRAI Times (Korea) Translated from Korean

    The AI industry’s interest has been focused on “When AGI will appear” However, the recent AI development will be sucked and until the view that will reach AGI in 2030, the discussion will lead to 'What will happen after AGI'. In this case…

  3. The most misunderstood graph in AI just got a major update
    USMIT Technology Review Biotech

    A new report from the AI evaluation nonprofit METR finds that some AI capabilities are continuing to improve along an exponential trajectory—but it also indicates that models aren’t quite as good at tasks that require judgment as they are…

  4. LightGBM vs. a Four-Week Average for Forecasting Bike-Share Rentals
    USTowards Data Science

    A 48-hour Capital Bikeshare experiment pits XGBoost, Ridge, SARIMA, and Prophet against simple baselines, with care for missing hours and leakage. The post LightGBM vs. a Four-Week Average for Forecasting Bike-Share Rentals appeared first…

  5. Clairic Hit 98% Accuracy on 72.8% of Questions via Confidence Scores
    USTowards Data Science

    With a 0.9 confidence cutoff, Cloudflare's System One model handles the easy cases and routes the rest to a bigger, slower LLM like Luna. The post Clairic Hit 98% Accuracy on 72.8% of Questions via Confidence Scores appeared first on…

  6. How to Prepare for a Credit Risk Internship Interview in Nine Steps
    USTowards Data Science

    Two years of interviewing master's students distilled into one logistic regression scoring model, from missing values and VIF to AUC and Gini. The post How to Prepare for a Credit Risk Internship Interview in Nine Steps appeared first on…

  7. The Download: inside the quest to make our minds immortal
    USMIT Technology Review Biotech

    This is today’s edition of The Download , our weekday newsletter that provides a daily dose of what’s going on in the world of technology. What we learned when our Roundtables explored the quest to make minds immortal To preserve our sense…

  8. Five Guardrails That Limit the Damage an AI Agent Can Do
    USTowards Data Science

    AI Agents for Beginners, Part three: strict Pydantic arguments, a tool allowlist, read-only SQLite, a step cap, and confirmed writes, all verified by five pytest tests in 0.68s. The post Five Guardrails That Limit the Damage an AI Agent…

  9. Combatting adversarial distillation
    USOpenAI Blog

    OpenAI’s layered approach to adversarial distillation combines model training, safeguards, detection, and enforcement to protect against capability extraction.

  10. Google, the use of the elemental rate table to AI model name.. . What's New
    KRAI Times (Korea) Translated from Korean

    Google has the name of the chemical element in the "Jeminai 4" and its unique clarity is a topic. Whenever the new model comes out, users are expecting the next element, and they are looking for a model selection and video that utilizes…

  11. OpenAI, 'GPT-6.2' responses get up to 40% discount for API fee
    KRAI Times (Korea) Translated from Korean

    OpenAI introduced a new fundraising method that applies API fees according to the response speed of AI model. When developers set time to wait for the response completion, they can save up to 40%, making it possible to reduce the cost…

  12. Meta, how AI remembers itself, improves... About Us
    KRAI Times (Korea) Translated from Korean

    Meta revealed the technology that AI agents will improve their memory management system based on work experience and reflect them in model learning. In recent years, OpenAI and Andropic announced the development and automation and…

  13. [Bulletin] Selection of Cellstar, Publishers Association and AI Learning Publishing Data Supply Cooperation
    KRAI Times (Korea) Translated from Korean

    ■ Select Star (CEO Kim Se-Fil) has signed a work agreement (MOU) to supply high-quality AI learning publishing data while protecting the rights of the Korean Journal Culture Association and the copyrighter and publisher. Both sides…

  14. Newro Fusion, AI Future Predictive Benchmark Co., Ltd.
    KRAI Times (Korea) Translated from Korean

    In the global AI future prediction benchmark, the AI specialist has recorded the first place in a major AI company such as Google Deep Mind and OpenAI, and Andropic. This is an explanation that demonstrates the competitiveness of AI…

  15. AI was first opened...A large number of malicious characters and spam calls are blocked before receiving
    KRAI Times (Korea) Translated from Korean

    Adress: No. 5 Yard, 7th Guanggu Street, Badaling Economic Development Zone, Yanqing District, Beijing, China The Korean Internet Promotion Agency (KISA) and Broadcast Media Communications Committee held ‘2026 spam data opening performance…

  16. U.S. Department of Commerce, AI Chip Export Authorization to China Military Corporation. . Home > About Us
    KRAI Times (Korea) Translated from Korean

    The fact that the U.S. Department of Commerce has been approved to export approximately 8 million AI chips from NVIDIA and AMD to China. More than 30% of the total approval amount is a relatively large view on the export permit screening…

  17. OpenAI, ChatGPT performance problem automatic diagnosis and check-in
    KRAI Times (Korea) Translated from Korean

    OpenAI has released an internal automation system that uses ‘codecs’ to find and modify performance degradation causes of ChatGPT. When performance issues occur, AI agent analyzes logs and codes, identify causes and writes the fixes, and…

  18. Google introduces "computer yuzu" function in Jeminai.. .Mac app direct control
    KRAI Times (Korea) Translated from Korean

    Google introduced the ‘Computer Use’ feature that directs computers to the ‘Jeminai’ app for Mac. Antropicy and OpenAI have been featured in 1~2 years ago, and Google has also expanded the existing developer technology to the general user…

  19. Development of a universal AI agent beyond Apple and Siri.. .App operation andcoding
    KRAI Times (Korea) Translated from Korean

    Apple has been developing a universal AI agent that can handle complex tasks autonomously than traditional ‘Siri AI’. We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will…

  20. MORE, LLM CUTTON Efficient Framework ‘TIDE’ released
    KRAI Times (Korea) Translated from Korean

    More (CEO Jo Gangwon) announced that it has revealed the ‘TIDE’ of self-improved abstract framework that increases the performance and efficiency of large-language model (LLM). TIDE was first announced at the HPC Conference ‘Supercomputing…

  21. AMD Lisa "Jenson Huang is a great leader.. . Nvidia Rhinolan and the street two"
    KRAI Times (Korea) Translated from Korean

    Lee Se-soo, CEO of AMD, said the position to put the streets in various fields surrounding Nvidia, as well as the excellent leader in the industry. In the event of a change in the competition, it is necessary to avoid direct mentions in…

  22. Nvidia negotiates acquisition of DLLM developer inception over 6700 billion
    KRAI Times (Korea) Translated from Korean

    Nvidia has been known to be a march negotiation to acquire at least $500 billion (approximately 6700 billion) for the development of DLLM. DLLM technology and core research workforces fast speed, and is full of strategy to strengthen its…

  23. Alababa, Nvidia Spy Scandal-related suspicious lawsuits
    KRAI Times (Korea) Translated from Korean

    Alababa has been a shareholder litigation in the U.S. for the reason why it has been lost to investors by hiding the relationship with the company Megaspeed, which is linked to the Chinese-related semiconductor exemption. It is a claim…

  24. AMD, a 8% recurring share in China's sanctions and deals.. . U.S. investigation potential raised
    KRAI Times (Korea) Translated from Korean

    AMD has maintained trading relations with Chinese companies, the United States’ sanction target, and about 8% of the shareholders’ report. The company’s company’s company’s company’s company’s company’s company’s company’s company’s…

  25. KT·LG Electronics Implementation of Yangjae Telecom System...Imiji·Cut Transfer·Received
    KRAI Times (Korea) Translated from Korean

    KT (CEO Park Yoon-young) announced that it implements the ‘QSDC’ system that transmits digital information to LG Electronics (CEO Ryu Jae-gil) and Yangjae status, and has succeeded in sending images and text data. QSDC is the next…

  26. Doorshot AI, US export regulator, US $200 million in size ENVIDIA chips lease contract cancellation
    KRAI Times (Korea) Translated from Korean

    Moonshot AI has canceled a contract of $500 million (about 2000 billion) to rent the latest AI chips in Nvidia through the Indonesia data center. U.S. announced a new regulation that restricts the use of high-tech AI chips for Chinese…

  27. 'GPT-6.2 Astra'·'Seminai 4 Neon', Water-proof Manufacture... LG
    KRAI Times (Korea) Translated from Korean

    OpenAI’s ‘GPT-6.2 Astra’ and Google’s ‘Jeminai 4 neon’ recorded a full range of subjects, including 17 subjects exploring in the 2027 class. Among the domestic models, LG AI researchers’ ‘K-Exam 2.5’ received the highest grade. Gyu-yu, a…

  28. Blue Orient and Contract to Build Google, Space AI Data Center.. .2029 Satellite Launch
    KRAI Times (Korea) Translated from Korean

    Google was selected as the launch partner of the Blue Duckjin, the space company of Jeep Bayjos to build a solar-based AI data center in the universe. In the early 2029, we plan to launch a series of satellites with AI semiconductors, and…

  29. When Should Deployed BCI Decoders Adapt? A Resource-Priced Ladder for Frozen EEG Trunks
    USarXiv cs.LG

    arXiv:2611.25315v1 Announce Type: cross Abstract: When should a deployed MI decoder spend adaptation resources-not only how to adapt under shift? We study a thin control layer outside a frozen encoder: a resource ladder (factory ->…

  30. ICARE: cross-domain text classification with incremental class-aware representation and distillation learning
    USarXiv cs.LG

    arXiv:2611.24959v1 Announce Type: cross Abstract: The rise of powerful, fine-tuned large language models (LLMs) necessitates robust natural language inference (NLI) systems. These NLI systems ideally train across various tasks…

  31. Sequential Anytime Valid Tests for Efficient Inference-Time Alignment
    USarXiv cs.LG

    arXiv:2611.26413v1 Announce Type: new Abstract: Inference-time alignment methods, such as Best-of-N (BoN), improve language model outputs by generating multiple candidates per prompt and selecting among them using a reward model. Much of…

  32. Evidence from a Multi-Regime Causal Discovery Benchmark: UCMT Algorithms, Comparative Performance, and Controlled Ablation Studies
    USarXiv cs.LG

    arXiv:2611.26383v1 Announce Type: new Abstract: We present an internal, pre-registered evaluation of the UCMT causal discovery algorithm family alongside established alternatives across heterogeneous benchmark regimes, complemented by a…

  33. Complete-SVD: A Functional Norm Approach to Low-Rank LLM Compression
    USarXiv cs.LG

    arXiv:2611.26059v1 Announce Type: new Abstract: Post-training low-rank compression of neural networks must not only approximate the original weights but also preserve the model's input-output behavior. Recent output-aware methods measure…

  34. Reinforcing Few-Step Diffusion Models via Shadow Posterior Matching
    USarXiv cs.LG

    arXiv:2611.25998v1 Announce Type: new Abstract: Few-step diffusion models are widely deployed for their efficiency. However, reinforcing them with black-box rewards, as is routine for multi-step models, faces two structural challenges…

  35. Interpretable Prototype-Conditioned Adaptation for Weather Forecasting
    USarXiv cs.LG

    arXiv:2611.25848v1 Announce Type: new Abstract: Data-driven weather forecasting requires distinguishing local states and their temporal evolution. However, learned patterns remain implicit in neural forecasters, making their influence…

  36. Disentangling Monitor Hardware Artifacts and Smoothing Lag for Shift-Robust Intraoperative EEG and Hemodynamic Risk Calibration
    USarXiv cs.LG

    arXiv:2611.25765v1 Announce Type: new Abstract: Commercial intraoperative electroencephalography (EEG) and blood pressure monitors introduce cyclic impedance pulses, smoothing delays, direct-current (DC) drift, and cuff holds that confound…

  37. Explainability of genotype embedding models through observed cohort paths
    USarXiv cs.LG

    arXiv:2611.25681v1 Announce Type: new Abstract: Deep learning models that combine genotypes with neuroimaging predict neurological diseases more accurately than either modality alone. Their clinical use, however, requires identifying which…

  38. PSTO: A PDE-structured transport operator with shared implicit updates for multiscale CDR systems
    USarXiv cs.LG

    arXiv:2611.25607v1 Announce Type: new Abstract: Multiscale convection-diffusion-reaction (CDR) systems couple spatial transport with local stiff reaction kinetics, making their numerical simulation computationally demanding. Conventional…

  39. From Data Alignment to Parameter Generation: Few-Shot Cross-City Traffic Prediction with Conditional Diffusion
    USarXiv cs.LG

    arXiv:2611.25592v1 Announce Type: new Abstract: Accurate mobile traffic forecasting is crucial for network service management, supporting proactive capacity provisioning, congestion mitigation, and energy-efficient operation. However, many…

  40. SynthDCB: World Distillation for Scalable Device Control Benchmarking
    USarXiv cs.AI

    arXiv:2611.18639v2 Announce Type: replace-cross Abstract: Scaling device-control evaluation requires scaling both interactive worlds and the tasks agents must solve. We introduce SynthDCB, a benchmark for environment awareness and device…

  41. Why Trust Your Agent? Empirical Security Gains from TRiSM-Guided Agentic Workflows in Healthcare
    USarXiv cs.AI

    arXiv:2606.28666v2 Announce Type: replace-cross Abstract: Agent-based AI has enabled the automation of tasks by exposing application tools and resources to large language models (LLMs). However, to improve scope and accuracy, agents are…

  42. Random coloured digraphs defined by a Markov logic network
    USarXiv cs.AI

    arXiv:2606.23715v2 Announce Type: replace-cross Abstract: A Markov Logic Network (MLN) $M$ is a probabilistic relational model used in Statistical Relational Artificial Intelligence for defining a probability distribution $P_D^M$ on the…

  43. Learning Which Evidence-Seeking Decisions Matter: Calibrated Counterfactual Replay for Language Agents
    USarXiv cs.AI

    arXiv:2606.05263v2 Announce Type: replace-cross Abstract: Language agents answer questions by searching, reading, and checking claims. A reward on the final answer says little about which of those decisions earned it, since an agent can…

  44. PhoneWorld: From Real-App Trajectories to Dynamic and Verifiable Environments for Phone-Use Agents
    USarXiv cs.AI

    arXiv:2605.29486v4 Announce Type: replace-cross Abstract: Real applications provide the training setting closest to phone-agent deployment, but are difficult to reset, scale safely, and verify programmatically. Static screenshots and…

  45. Periodic and Harmonic Periodic RoPE for Length Extrapolation in LLMs
    USarXiv cs.AI

    arXiv:2605.27980v2 Announce Type: replace-cross Abstract: The ability to process long contexts is crucial for large language models (LLMs) to perform long-horizon tasks. Rotary position embedding (RoPE) gives every token a new rotation…

  46. MINTEval: Evaluating Memory under Multi-Target Interference in Long-Horizon Agent Systems
    USarXiv cs.AI

    arXiv:2605.18565v3 Announce Type: replace-cross Abstract: Real-world agents operate over long and evolving horizons, where information is repeatedly updated and may interfere across memories, requiring accurate recall and aggregated…

  47. Authors' Response to AI-Based Reviews: A Mixed-Methods Pilot Study
    USarXiv cs.AI

    arXiv:2605.16623v2 Announce Type: replace-cross Abstract: Background: Large language models are increasingly used to assist scholarly peer review, but there is limited evidence about how authors judge their feedback and use it to revise…

  48. VLRS-Bench: A Vision-Language Reasoning Benchmark for Remote Sensing
    USarXiv cs.AI

    arXiv:2602.07045v3 Announce Type: replace-cross Abstract: Recent advancements in Multimodal Large Language Models (MLLMs) have enabled complex reasoning. However, existing remote sensing (RS) benchmarks remain heavily biased toward…

  49. LLM-Based Threat Detection and Prevention Framework for IoT Ecosystems
    USarXiv cs.AI

    arXiv:2505.00240v3 Announce Type: replace-cross Abstract: The rapid expansion of the Internet of Things (IoT) has introduced complex and evolving cybersecurity challenges. This paper presents a Large Language Model (LLM)-based framework…

  50. LimiX-2: A Contextual Mechanism Network Towards General Structured-Data Intelligence
    USarXiv cs.AI

    arXiv:2609.17488v2 Announce Type: replace Abstract: We introduce LimiX-2, a new model in the LimiX family, developed through model and data scaling guided by our previously established scaling laws. LimiX-2 adopts the Contextual Mechanism…

  51. CHORUS: Coverage-Guided Agentic RL and Policy Consolidation for Testbench Stimulus Generation
    USarXiv cs.AI

    arXiv:2608.10090v2 Announce Type: replace Abstract: Testbenches written by large language models can compile and run while leaving entire datapaths idle. We present CHORUS, the first framework to combine coverage-guided agentic…

  52. Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
    USarXiv cs.AI

    arXiv:2607.22583v4 Announce Type: replace Abstract: Large Language Models (LLMs) have achieved widespread adoption because of their strong reasoning and query-response capabilities. However, deploying them in embedded and edge computing…

  53. AI GameStore: Scalable, Open-Ended Evaluation of Machine General Intelligence with Human Games
    USarXiv cs.AI

    arXiv:2602.17594v2 Announce Type: replace Abstract: Rigorously evaluating machine intelligence against the broad spectrum of human general intelligence has become increasingly important and challenging in this era of rapid technological…

  54. Multi-institutional grading of Barrett's esophagus dysplasia without stain normalization using a pathology foundation model and multiple instance learning
    USarXiv cs.AI

    arXiv:2611.26562v1 Announce Type: cross Abstract: Barrett's esophagus dysplasia grading suffers substantial inter-observer variability, particularly for low-grade dysplasia, complicating clinical decision-making. We conducted a…

  55. Cost-Aware Agentic Feature Optimization System for Fleet-Scale Ranking Models
    USarXiv cs.AI

    arXiv:2611.26531v1 Announce Type: cross Abstract: Large Web platforms operate large fleets of ranking models, each selecting features from a shared pool of tens of thousands. We present an end-to-end agentic system that optimizes features…

  56. Asuka-FM: A System-Architectural Co-Design Framework for Efficient Text-to-Image Pretraining on Consumer Hardware
    USarXiv cs.AI

    arXiv:2611.26437v1 Announce Type: cross Abstract: Frontier text-to-image (T2I) foundation models predominantly rely on multi-billion parameter Diffusion Transformers (DiTs) pretrained across massive compute clusters, creating an…

  57. Flow Matching from Non-Gaussian Sources: A Theory of Efficient Sampling
    USarXiv cs.AI

    arXiv:2611.26407v1 Announce Type: cross Abstract: Flow matching is a versatile framework for generative modeling that transports a source distribution to a target data distribution. A key advantage is its flexibility in the choice of…

  58. Beam Model Technical Report
    USarXiv cs.AI

    arXiv:2611.26358v1 Announce Type: cross Abstract: Beam is a sparse Mixture-of-Experts (MoE) model with 501.3B total parameters and 23.4B active parameters, specifically built and optimized for coding, reasoning, and agentic workloads. Beam…

  59. SkillBridge: Cost-Aware Diagnosis and Repair of Cross-OS Compatibility Defects in Agent Skills
    USarXiv cs.AI

    arXiv:2611.26256v1 Announce Type: cross Abstract: LLM agents increasingly reuse executable capabilities through skills, but a skill that works on one OS may fail on another because OS assumptions are spread across guide text, scripts…

  60. The Stability Gap in Continual Reinforcement Learning
    USarXiv cs.AI

    arXiv:2611.26252v1 Announce Type: cross Abstract: Continual reinforcement learning (CRL) aims to enable agents to adapt to changing tasks while preserving previously acquired capabilities. Evaluation often relies on final aggregate metrics…

  61. FastRoverVLA: Fast Vision-Language-Action Execution for Off-Road Autonomous Driving
    USarXiv cs.AI

    arXiv:2611.26242v1 Announce Type: cross Abstract: Off-road vision-language-action (VLA) models generate scene-conditioned plans, but frequent trajectory updates can repeatedly invoke expensive multimodal inference. Reusing a prediction…

  62. Sightline: Calibrated Data-Flow Summaries for LLM-Integrated Applications
    USarXiv cs.AI

    arXiv:2611.26221v1 Announce Type: cross Abstract: Programs increasingly call large language models (LLMs), filling prompt placeholders with program values and consuming the output. Such a call crosses the boundary between natural language…

  63. GenSPARQL: A Generative Extension of SPARQL for Incomplete Knowledge Graphs
    USarXiv cs.AI

    arXiv:2611.26062v1 Announce Type: cross Abstract: Knowledge graphs on the Web are incomplete, and SPARQL answers a query only from the stored triples. Large language models (LLMs) can predict missing facts, but an LLM returns names instead…

  64. Web search in Latin America: language models use less local and diverse news than search engines
    USarXiv cs.AI

    arXiv:2611.25955v1 Announce Type: cross Abstract: AI platforms are becoming a major gateway to news and current events. This shift risks amplifying existing geographical biases that favour narratives from wealthier countries over those…

  65. On the trajectories of the ID algorithm
    USarXiv cs.AI

    arXiv:2611.25877v1 Announce Type: cross Abstract: The ID algorithm decides whether an interventional distribution can be computed from observational data and, when it can, returns a formula in which products, ratios and summations of…

  66. Collision-Free Is Not Enough: Passage-Conditioned Navigation for Wheeled-Legged Robots
    USarXiv cs.AI

    arXiv:2611.25788v1 Announce Type: cross Abstract: A collision-free detour can still bring a legged robot to the next stair or ledge a few centimetres or degrees off, and a fixed locomotion skill, factory or learned, can stall or fall…

  67. Characterizing and Detecting Unfaithful Work Reports by Coding Agents
    USarXiv cs.AI

    arXiv:2611.25760v1 Announce Type: cross Abstract: Coding agents increasingly carry out long-horizon development tasks, autonomously planning and executing multi-step actions that span files, tools, and sessions. Because developers rarely…

  68. An LLM-Driven Traffic-Control xApp for Dynamic 5G RAN Intents
    USarXiv cs.AI

    arXiv:2611.25736v1 Announce Type: cross Abstract: Programmable Radio Access Networks (RANs) enable runtime deployment of traffic-control functions, but selecting the appropriate configuration remains challenging as network conditions and…

  69. From Reliance to Overreliance: Taxonomizing and Detecting Relational Risks in AI Companionship
    USarXiv cs.AI

    arXiv:2611.25734v1 Announce Type: cross Abstract: Companion AIs are increasingly used for emotional support, yet distinguishing meaningful engagement from harmful reliance remains difficult. We introduce the Taxonomy of Relational Risks…

  70. Verifying Before Enhancing: Target Evidence Chain Verification for Infrared Small Target Detection
    USarXiv cs.AI

    arXiv:2611.25724v1 Announce Type: cross Abstract: Infrared small target detection is essential for early warning, surveillance, remote sensing, and space-based target perception. However, infrared small targets are typically tiny, weakly…

  71. SteerTrace: Tracing How Target Attacks Steer World Action Models
    USarXiv cs.AI

    arXiv:2611.25719v1 Announce Type: cross Abstract: World action models (WAMs) predict future videos and actions with one backbone, and their imagined future is often seen as a safety signal. We introduce SteerTrace, a targeted universal…

  72. Tether: Verified Stateful Policy Engine for System-Level Monitoring of Agents
    USarXiv cs.AI

    arXiv:2611.25599v1 Announce Type: cross Abstract: Autonomous AI agents are increasingly deployed for long-running tasks that execute arbitrary shell commands or programs. However, these actions can result in unauthorized network requests…

  73. An Explainable and Scalable Hierarchical Tensor Framework for Integrating Multi-Omics and Digital Pathomics
    USarXiv cs.AI

    arXiv:2611.25584v1 Announce Type: cross Abstract: Background: Precision oncology requires integration of heterogeneous molecular and pathology data, yet many fusion models offer limited biologically grounded interpretability and struggle…

  74. AutoRAGBench: A Benchmark for LLM Agents Automating RAG System Development
    USarXiv cs.AI

    arXiv:2611.25582v1 Announce Type: cross Abstract: Retrieval-augmented generation (RAG) is a well-established paradigm for augmenting large language models (LLMs) with information retrieved from external corpora. However, developing…

  75. FlashX: Scaling Robotic Experience Reuse for Embodied Agents with CBA-Based NAND Flash
    USarXiv cs.AI

    arXiv:2611.25580v1 Announce Type: cross Abstract: Embodied agents are emerging as a promising paradigm for complex robotic tasks, leveraging Vision-Language-Action (VLA) and World Action Models (WAMs) to generate actions through Diffusion…

  76. The Coverage Selection Gap: Ranking, Not Generation, Limits LLM Rubric Construction
    USarXiv cs.AI

    arXiv:2611.25557v1 Announce Type: cross Abstract: Task-specific rubrics are increasingly used to evaluate open-ended language-model outputs, yet constructing comprehensive evaluation criteria from complex source documents remains…

  77. MATE: Mobile Agent Trustworthy Execution for Safe and Efficient Mobile Control
    USarXiv cs.AI

    arXiv:2611.25526v1 Announce Type: cross Abstract: Mobile agents on personal smartphones can affect private data, payments, settings, and the foreground screen. We present MATE (Mobile Agent Trustworthy Execution), an execution-layer system…

  78. AIstatLab at BraTS-GoAT 2026 Task 3: Quality-Controlled Consensus Pseudo-Labeling and Dense Expert Adaptation
    USarXiv cs.AI

    arXiv:2611.25477v1 Announce Type: cross Abstract: BraTS-GoAT Task 3 evaluates brain-tumor segmentation across heterogeneous entities using shared ET, TC, and WT regions. We present a 3D ResidualEncoderUNet trained with quality-controlled…

  79. A four-level executable validation protocol for evaluating LLM-generated topology specifications of physics-based HVAC digital twins
    USarXiv cs.AI

    arXiv:2611.25471v1 Announce Type: cross Abstract: Physics-based digital twins of air conditioning, a principal energy consumer in manufacturing and buildings, can be generated automatically from a declarative specification of its…

  80. Demonstration-Free Skill Composition for Navigation and Manipulation
    USarXiv cs.AI

    arXiv:2611.25462v1 Announce Type: cross Abstract: Hierarchical reinforcement learning enables sequential tasks to be represented as compositions of separately trained skills. This paper evaluates demonstration-free hierarchical pipelines…

  81. Bilateral Evidence Selection and Reasoning for Mammography Vision--Language Models
    USarXiv cs.AI

    arXiv:2611.25418v1 Announce Type: cross Abstract: Medical vision--language models (VLMs) often need to identify small, subtle lesions within high-resolution images containing extensive normal anatomy and redundant visual information…

  82. SGLayout: Schema-Guided Closed-Loop Refinement for LLM-Driven Parameterized Layout Generation
    USarXiv cs.AI

    arXiv:2611.25410v1 Announce Type: cross Abstract: Large Language Models (LLMs) have shown strong potential in parameterized layout generation via tool-augmented code synthesis. However, LLMs lack spatial awareness, making one-shot…

  83. FPGA-GEN: Target-Aware RTL Generation via HLS
    USarXiv cs.AI

    arXiv:2611.25351v1 Announce Type: cross Abstract: Successful C/C++ compilation does not guarantee correct or efficient FPGA hardware after high-level synthesis (HLS). FPGA-GEN is an agentic HLS-C generation workflow that encodes…

  84. LLMs Are Already Confidence-Aware Zero-Shot Classifiers
    USarXiv cs.AI

    arXiv:2611.25289v1 Announce Type: cross Abstract: Zero-shot large language models (LLMs) are increasingly used as text classifiers, but their labels are only useful when we know to what extent we can trust them. We compare black-box…

  85. When Objectives and Evaluators Disagree: Evaluating Transfer for Post-hoc Demographic Framing Mitigation
    USarXiv cs.AI

    arXiv:2611.25258v1 Announce Type: cross Abstract: Post-hoc rewriting is used to reduce demographic disparities in LLM-generated text when the source model is inaccessible. A rewrite can look successful merely because it improved its own…

  86. Plan your Self-Adaptation! Ontology-Mediated Task and Architecture Co-adaptation Planning for Robots
    USarXiv cs.AI

    arXiv:2611.25254v1 Announce Type: cross Abstract: Autonomous robots may need to adapt both their task plans and architectural configurations as operating conditions change. These decisions are interdependent: architectural configurations…

  87. Some Memories Must Fade: Persistence and Decay of User Memories in Model Parameters
    USarXiv cs.AI

    arXiv:2611.25235v1 Announce Type: cross Abstract: Parametric memory allows Large Language Models to internalize knowledge directly into their weights, offering a potentially powerful alternative to retrieval-based memory (RAG). However…

  88. VG-Steer: Visual-Geometric Policy Steering from In-Context Aligned Abstractions
    USarXiv cs.AI

    arXiv:2611.25199v1 Announce Type: cross Abstract: Adapting robot policies to out-of-distribution (OOD) settings with limited real-world data remains a key bottleneck to generalist robots. The multimodality and scarcity of robotics data…

  89. MADJEPA: Enhanced Multi-Stage Joint-Embedding Predictive Architecture for Medical Anomaly Detection
    USarXiv cs.AI

    arXiv:2611.25194v1 Announce Type: cross Abstract: Anomaly detection in medical images is a complex task, requiring the development of a robust model using only healthy patient data. Without access to labeled data, modern approaches…

  90. Before the Choice: An Evidential Framework for Studying Prospective Representations in Language Models
    USarXiv cs.AI

    arXiv:2611.25170v1 Announce Type: cross Abstract: An observable decision reveals the selected action but does not determine which alternatives were internally represented. Drawing on prospection research and hippocampal forward-sweep…

  91. CRAFT: Constrained Route planning with Adaptive Framework and Template-based code generation
    USarXiv cs.AI

    arXiv:2611.25128v1 Announce Type: cross Abstract: Everyday routing requests are rarely a point-to-point shortest path. A driver may need to visit stops, sometimes in a fixed order, under time budgets or distance limits, and perhaps…

  92. MAWAM: A Multi-Agent World-Action Model for Coordinated Manipulation
    USarXiv cs.AI

    arXiv:2611.25105v1 Announce Type: cross Abstract: Every move in a robot team changes the world in which its teammates must act. Coordinated manipulation therefore calls for predicting the team's actions together with the physical future…

  93. FinVeriQA: Financial Research Agents for Question Answering: Routing, Retrieval, MCP Tool Use, and Provenance-Aware Evidence Verification
    USarXiv cs.AI

    arXiv:2611.25071v1 Announce Type: cross Abstract: Large language models (LLMs) write fluent financial analyses but miscompute ratios, recall stale prices, and route questions to the wrong tool with high confidence. We ask which component…

  94. DVBench: End-to-End Benchmark for Real-World UVM Design Verification Agents
    USarXiv cs.AI

    arXiv:2611.25066v1 Announce Type: cross Abstract: Design verification (DV) remains a major bottleneck in hardware development, with large-scale verification using the Universal Verification Methodology (UVM) being highly labor-intensive…

  95. Near-Quadratic Sample Compression for All VC Classes
    USarXiv cs.AI

    arXiv:2611.25011v1 Announce Type: cross Abstract: The sample compression conjecture is a longstanding open problem in learning theory, originating in the work of Littlestone and Warmuth in the 1980s. At its heart is a connection between…

  96. DL101 The Representation Capability of Neural Network Models
    USarXiv cs.AI

    arXiv:2611.24980v1 Announce Type: cross Abstract: Universal approximation theorems (Cybenko, Funahashi, Hornik & Leshno) provide the mathematical justification for why neural networks are powerful enough to model complex, nonlinear…

  97. Trace2ST: Recovering PLC Structured Text from I/O Traces with LLM Generation and Verification
    USarXiv cs.AI

    arXiv:2611.24979v1 Announce Type: cross Abstract: Programmable logic controllers (PLCs), the industrial computers that operate power and process plants, often run programs whose source code is no longer available; however, every input and…

  98. No-Harm-Jury A Jury-Based Multi-Dimensional Harm Evaluation Framework for Medical AI Responses
    USarXiv cs.AI

    arXiv:2611.24972v1 Announce Type: cross Abstract: We present No-Harm-Jury, a jury-based automated evaluation framework that scores AI-generated medical responses across seven conceptually distinct harm dimensions using a panel of five…

  99. Probing the "Conscious" Mind: Selective Access and Global Workspace in Vision-Language Models
    USarXiv cs.AI

    arXiv:2611.26590v1 Announce Type: new Abstract: Vision-language models (VLMs) integrate visual and textual information, but their outputs do not reveal internal concept representations. Global workspace theory (GWT) proposes that…

  100. A CPU Inside an LLM: Zero-Gradient Compilation of Exact Circuits into Pretrained Weights
    USarXiv cs.AI

    arXiv:2611.26450v1 Announce Type: new Abstract: Pretrained language models (LMs) remain unreliable on exact multi-digit arithmetic without gradient fine-tuning or external tool interpreters. We introduce closed-form, zero-gradient…

  101. Reasoning-Trace Analysis Using Large Language Models for Interpretable Q-Matrix Construction
    USarXiv cs.AI

    arXiv:2611.26422v1 Announce Type: new Abstract: Large language models (LLMs) can generate Q-matrices for diagnostic classification models that approximate expert specifications, but evaluations have focused on the final matrices, leaving…

  102. How Much Should an Agent Remember? Deployment Lessons from Self-Distilled Memory for LLM Agents
    USarXiv cs.AI

    arXiv:2611.26361v1 Announce Type: new Abstract: LLM agents on long-horizon tasks fail in recurring, procedural ways. They mis-paginate an API, resolve the wrong entity, or answer when no answer was asked. We built a pipeline that lets an…

  103. Context as a Resource
    USarXiv cs.AI

    arXiv:2611.26346v1 Announce Type: new Abstract: Managing concurrent tasks across separate agent sessions quickly exhausts human attention, motivating always-on personal agents behind a single continuous interface. Yet when every context…

  104. System-One RSI: Building Muscle Memory for Interactive Agents through Guided Practice
    USarXiv cs.AI

    arXiv:2611.26007v1 Announce Type: new Abstract: Reasoning models provide strong multimodal task understanding but struggle with sub-second control because deliberation increases decision latency. Fast decision models can meet the response…

  105. Meta-RTL: Evolving Agent Harnesses for RTL Optimization
    USarXiv cs.AI

    arXiv:2611.25891v1 Announce Type: new Abstract: Large language models enable agents to optimize RTL using EDA feedback, yet their executable search procedures remain largely hand- engineered. We present Meta-RTL, an EDA-grounded framework…

  106. ADAPT: Adaptive Decision and Abstention with Pool-Size-Aware Thresholds for Service Routing
    USarXiv cs.AI

    arXiv:2611.25754v1 Announce Type: new Abstract: Retrieval-based classifiers efficiently route queries to specialised services, but they typically select a candidate even when no suitable service exists. We introduce Adaptive Decision and…

  107. How Feedback-Free Self-Evolution Reshapes Agent Skills: A Systematic Study
    USarXiv cs.AI

    arXiv:2611.25534v1 Announce Type: new Abstract: Agent skills are documents in which experts write down procedural knowledge, and models increasingly maintain them by rewriting them without feedback. Such rewriting could adapt a skill to…

  108. LM-based Personalized Travel Planner
    USarXiv cs.AI

    arXiv:2611.25531v1 Announce Type: new Abstract: Effective travel planning requires handling multiple information sources, managing constraints, and tailoring recommendations to user preferences. While LLMs have introduced dynamic…

  109. Coordinated Use of Large Language Models in Academic Research
    USarXiv cs.AI

    arXiv:2611.25502v1 Announce Type: new Abstract: This study proposes a methodological framework for the coordinated use of multiple large language models in academic research. Its central argument is that involving more models does not in…

  110. Why Do LLMs Help with Scientific Misconduct Requests and What to Do About It?: Increasing the Trustworthiness of AI for Science
    USarXiv cs.AI

    arXiv:2611.25492v1 Announce Type: new Abstract: Large language models (LLMs) support scientific research, but some of that assistance can also undermine research integrity. Existing safety evaluations in science often measure whether…

  111. Stop Going in Circles:Loop-Triggered Routing for System 1 Decision Agents
    USarXiv cs.AI

    arXiv:2611.25363v1 Announce Type: new Abstract: System 1 decision models (represented by Jev) provide fast, low-cost action selection, but existing agent integrations do not reliably translate these local advantages into task completion…

  112. Pandora's Toolbox Or Why Agent Safety Cannot Be Verified
    USarXiv cs.AI

    arXiv:2611.25325v1 Announce Type: new Abstract: Present artificial intelligence agents are constructed by means of a coupling between a fixed language model with a general computing device. We show that this coupling makes safety…

  113. A Neurosymbolic Multimodal Framework for 30-Day Diabetic Readmission Prediction: Integrating Structured EHR Data, Dense Clinical Embeddings, and Knowledge-Grounded Symbolic Extraction
    USarXiv cs.AI

    arXiv:2611.25236v1 Announce Type: new Abstract: Unplanned 30-day readmissions among patients with diabetes mellitus impose a substantial clinical and economic burden, yet most predictive models use only structured electronic health record…

  114. When AI Analysts Disagree: Agentic Multiverse Analysis and the Robustness of Empirical Research
    USarXiv cs.AI

    arXiv:2611.25198v1 Announce Type: new Abstract: Every empirical analysis on observational data rests on dozens of defensible analytical choices ("forks") that its readers cannot see. We develop an agentic multiverse analysis (AMA)…

  115. SailSearch: Agent-Assisted Discovery in Combinatorial Optimization
    USarXiv cs.AI

    arXiv:2611.25087v1 Announce Type: new Abstract: We present SailSearch, a human-guided workflow for algorithmic research with two language-model agents, and report results on Market Split and maximum independent set (MIS). For Market Split…

  116. How many sorted projections determine any n points in k dimensions?
    USarXiv cs.LG

    arXiv:2611.06470v3 Announce Type: replace-cross Abstract: Let $d_1,\dots,d_m$ be directions in $\mathbb{R}^k$, where $k\ge 2$. A multiset $X$ of $n$ points of $\mathbb{R}^k$ projects to the $m$ multisets of numbers $\{\{x\cdot d_j : x\in…

  117. Privacy-Utility Tradeoffs with Quantum Local Differential Privacy
    USarXiv cs.LG

    arXiv:2602.10510v2 Announce Type: replace-cross Abstract: When sensitive information is encoded in data, it is important to ensure the privacy of information when attempting to learn useful information from the data. There is a natural…

  118. Necessary, Not Sufficient, and No Longer Available: Grader Non-Determinism After the Withdrawal of Sampling Controls
    USarXiv cs.LG

    arXiv:2606.26185v3 Announce Type: replace Abstract: LLM-as-judge ("grader") components are now standard in evaluation harnesses, including safety evaluations where a pass/fail verdict may gate downstream deployment decisions. Earlier work…

  119. Benchmarking Robust Aggregation in Decentralized Gradient Marketplaces
    USarXiv cs.LG

    arXiv:2509.05833v2 Announce Type: replace Abstract: Institutions that cannot centralize or sell raw data increasingly exchange model updates instead: cross-silo federated learning is already in production among hospital networks and…

  120. Almost Sure Convergence Rates and Concentration of Stochastic Approximation and Reinforcement Learning with Markovian Noise
    USarXiv cs.LG

    arXiv:2411.13711v2 Announce Type: replace Abstract: This paper establishes, to our knowledge, the first almost sure convergence rate and the first maximal concentration bound with sub-Weibull tails for general contractive stochastic…

  121. Phantom of the Masquerade: Graph Neural Network-Based Detection of Masquerade Attacks in Controller Area Networks
    USarXiv cs.LG

    arXiv:2611.26094v1 Announce Type: cross Abstract: The lack of built-in security in controller area networks (CANs) exposes modern vehicles and industrial systems to masquerade attacks that can compromise safety and reliability. These…

  122. AdaptReg: Skip-level FiLM Adapter for Task-Specific Adaptation of a Frozen Registration Foundation Model
    USarXiv cs.LG

    arXiv:2611.25761v1 Announce Type: cross Abstract: Longitudinal PSMA-PET/CT registration is a key step in assessing treatment response in prostate cancer, and it requires more than generic alignment accuracy: the deformation must preserve…

  123. CGM: Chaos-Guided Graph Attention and Expert Routing for Autism Stereotypy Detection
    USarXiv cs.LG

    arXiv:2611.25730v1 Announce Type: cross Abstract: Skeleton graph attention for autism video analysis weights joints by content alone, giving an erratic wrist the same standing as a still one. Stereotypy, however, is defined by how…

  124. AdsNet: A Graph Neural Network for Direct Adsorption Energy Prediction
    USarXiv cs.LG

    arXiv:2611.25646v1 Announce Type: cross Abstract: Fast and accurate prediction of adsorption energies is critical for high-throughput computational catalyst screening, existing methods compromise either on evaluation speed (full-graph…

  125. Dynamics of Protected State Evolution
    USarXiv cs.LG

    arXiv:2611.25541v1 Announce Type: cross Abstract: States with identical protected observations can admit different future interventions. Each transition specifies its successor, response increment, required initial reserve and net…

  126. ActMem: From Latent Memory Reasoning to Grounded Robotic Action
    USarXiv cs.LG

    arXiv:2611.25461v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models have shown promising results in real-world long-horizon manipulation, where memory plays a crucial role. However, existing memory-augmented VLAs either…

  127. Instance-Optimal Moment Estimation in the Add-Remove Model of Differential Privacy
    USarXiv cs.LG

    arXiv:2611.25425v1 Announce Type: cross Abstract: In this paper, we study the problem of estimating empirical means, variances, and covariances under pure differential privacy in the add-remove model. While existing mechanisms achieve…

  128. Coordinate Basis Linear Bandits with Sublinear Regret over Low-Rank Clusters
    USarXiv cs.LG

    arXiv:2611.25424v1 Announce Type: new Abstract: We study the problem of computational efficiency for linear bandits in high-dimensional settings with a finite arm set. In linear bandits, the increase in the dimension $d$ of the feature…

  129. Antom 3-in-1 Transformer: A Payment Foundation Model
    USarXiv cs.LG

    arXiv:2611.25403v1 Announce Type: new Abstract: We report the development of Antom 3-in-1 Transformer, a payment foundation model that integrates sequence, tabular, and graph representations for payment risk detection. The model contains…

  130. TuringSpace: Grounding Spatial Intelligence in Action with Interactive CAPTCHAs
    USarXiv cs.LG

    arXiv:2611.25387v1 Announce Type: new Abstract: Humans act in the visual world by inferring the 3D layout behind a 2D view, predicting how an action changes it, and correcting themselves after seeing the result. Can multimodal large…

  131. Now You See Me Now You Don't: A Framework To Detect and Reduce Proxy Capacity
    USarXiv cs.LG

    arXiv:2611.25327v1 Announce Type: new Abstract: In regulated domains such as lending, the two most salient fairness concepts are disparate impact and disparate treatment: disparate impact is generally defined by differential rates of…

  132. Prescribed Bayesian Efficiency under Lifetime Adjustment Costs
    USarXiv cs.LG

    arXiv:2611.25311v1 Announce Type: new Abstract: A Bayesian learner may retain an issued action while updating its posterior, trading predictive loss against adjustment charges. We assume population curvature, boundary control, an…

  133. Federated Learning of Car Aerodynamics Surrogates with Spectral Optimizers
    USarXiv cs.LG

    arXiv:2611.25300v1 Announce Type: new Abstract: Neural surrogates predict the aerodynamic forces on a car in seconds, compared with hours for a CFD simulation, but training them requires thousands of CFD simulations of different car…

  134. Cross-Question Reliable Reinforcement Learning
    USarXiv cs.LG

    arXiv:2611.25292v1 Announce Type: new Abstract: AI systems operating near the limits of their capabilities can maximize utility while minimizing the risk of error by abstaining or deferring questions they are not confident about to a human…

  135. DS3M: Deep structured state space models for nonlinear system identification on modern hardware
    USarXiv cs.LG

    arXiv:2611.25274v1 Announce Type: new Abstract: Neural state-space identification learns dynamical models from measured inputs and outputs. SUBNET combines this task with an encoder that estimates initial states from short measurement…

  136. Moving Generality from Model to Modeler: LLM Agents for PHM on a Deployment-Style Benchmark
    USarXiv cs.LG

    arXiv:2611.25252v1 Announce Type: new Abstract: Coding agents built on large language models (LLMs) now help develop many models published as general solutions. Should generality instead reside in the agent that builds a model for each new…

  137. Learning to Read Events for Exact State Tracking with Frozen Language Models
    USarXiv cs.LG

    arXiv:2611.25211v1 Announce Type: new Abstract: Long event streams require identifying each change and applying the changes in order. We train a selector on a frozen language model to read each transition. With API-only access, we ask the…

  138. {\phi}TD: Distributional Reinforcement Learning using Characteristic Functions
    USarXiv cs.LG

    arXiv:2611.25208v1 Announce Type: new Abstract: Distributional Reinforcement Learning methods aim to learn the entire distribution of returns, yet current algorithmic approaches are limited to a narrow class of distributions. We propose…

  139. PDE-State Conditioning for Learned Optimization Using Physics-Informed Neural Networks
    USarXiv cs.LG

    arXiv:2611.25185v1 Announce Type: new Abstract: Physics-informed neural network (PINN) optimization depends on global properties of the underlying PDE and how the physics residual evolves across the domain. Standard optimizers such as Adam…

  140. Long-Range Video Presentation Attack Detection: A Three-Class Benchmark Across 1-100 m and Two Sensor Types
    USarXiv cs.LG

    arXiv:2611.25108v1 Announce Type: new Abstract: Face presentation attack detection (PAD) systems deployed in border control and airport surveillance are constrained by a critical assumption: subjects are captured at close range, typically…

  141. HapBO: Harnessed $\alpha$-$\pi$ Bayesian Optimization via Expert Natural Language Priors
    USarXiv cs.LG

    arXiv:2611.25100v1 Announce Type: new Abstract: Bayesian Optimization (BO) can be substantially more sample-efficient when informed by prior knowledge about the problem. In practice, however, such knowledge is often available only in…

  142. Watch Where You Linearize: Predicting and Correcting Model-Merging Interference
    USarXiv cs.LG

    arXiv:2611.25000v1 Announce Type: new Abstract: Model merging combines separately fine-tuned models by adding their task vectors to a shared backbone. Interference is typically assessed only after constructing and evaluating a candidate…

  143. Formalizing Formal Semantics: The Formal Atlas
    USarXiv cs.CL

    arXiv:2611.19509v2 Announce Type: replace Abstract: In this paper we present the Formal Atlas, a system for the study of natural language semantics using proof assistants. Our aim is to formalize semantic accounts, check their predictions…

  144. MedPsy: State-of-the-Art Small Medical Language Models for Efficient Edge Deployment
    USarXiv cs.CL

    arXiv:2611.02011v2 Announce Type: replace Abstract: Medical language models promise expert clinical support, but the strongest open-source LLMs remain too large for private edge-device deployment, while compact medical models still trail…

  145. HEDGE: Gated Hypergraph Memory for Scalable Multi-Agent Debate
    USarXiv cs.CL

    arXiv:2610.38902v2 Announce Type: replace Abstract: Multi-agent debate systems require agents to retain and revisit arguments across multiple rounds, yet limited context windows and unstructured memories make relevant information…

  146. From transcription to semantic corpus analysis: unsupervised learning of sentence representations for ancient languages
    USarXiv cs.CL

    arXiv:2607.24542v2 Announce Type: replace Abstract: Automatic Text Recognition (ATR) now supplies digital humanities with large volumes of unstruc- tured, heterogeneous, and often noisy text in ancient languages. Downstream semantic…

  147. Don't Lose Focus: Activation Steering via Key-Orthogonal Projections
    USarXiv cs.CL

    arXiv:2605.06342v2 Announce Type: replace Abstract: Activation steering controls LLM behaviour towards target behaviour by intervening in internal representations, yet it often degrades reasoning and retrieval performance. We argue that a…

  148. A minimal probe of task modeling in language models
    USarXiv cs.CL

    arXiv:2605.02028v3 Announce Type: replace Abstract: Language models read long contexts and score well on knowledge and reasoning benchmarks, but neither shows whether a model can do work whose next step depends on everything that came…

  149. DEO: Training-Free Direct Embedding Optimization for Negation-Aware Retrieval
    USarXiv cs.CL

    arXiv:2603.09185v2 Announce Type: replace Abstract: Recent advances in Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) have enabled diverse retrieval methods. However, existing retrieval methods often fail to…

  150. BADAS-3.0: Real-Time Driving Perception and Planning from One Camera
    USarXiv cs.CL

    arXiv:2611.26294v1 Announce Type: cross Abstract: We ask how far a driving model can get with one camera and real-time latency. BADAS-3.0 is a 1B-parameter video encoder with four small heads. From two seconds of one forward camera, it…