AI news archive.
10,660 articles filed under Research, newest first, from the outlets listed on the sources page.
- arXiv cs.AIresearchGenerative AI-Based Data Augmentation for Oral Lesion Classification: The PhotoMOCI Dataset and Benchmark
arXiv:2609.35226v1 Announce Type: cross Abstract: Early detection of oral cancer via photographic imaging presents a promising avenue for large-scale oral cavity screening. However, the development of robust deep learnin…
- arXiv cs.AIresearchToken-Disentangled Latent Test-Time Scaling for Vision-Language Reasoning
arXiv:2609.35228v1 Announce Type: cross Abstract: Latent test-time scaling improves reasoning by refining hidden states during inference, but existing methods typically apply a single scalar reward to all editable latent…
- arXiv cs.AIresearchSpatial Grafting: Grounding 3D Features for Flow-Matching Robot Policies
arXiv:2609.35249v1 Announce Type: cross Abstract: Pretrained robot manipulation policies such as vision-language-action models (VLAs) or world-action models (WAMs) leave interaction-relevant metric geometry implicit. Rec…
- arXiv cs.AIresearchWavePP: High-Throughput Pipeline Parallel LLM Prefill under Prefix Reuse
arXiv:2609.35263v1 Announce Type: cross Abstract: Pipeline parallelism can improve prefill throughput by processing multiple request chunks concurrently across different stages of the model. However, keeping the pipeline…
- arXiv cs.AIresearcheval-unlearn: Benchmarking unlearning in Text-to-Image Diffusion Models
arXiv:2609.35269v1 Announce Type: cross Abstract: The rising number of concept unlearning techniques for text-to-image (T2I) diffusion models has produced a fragmented evaluation landscape. Methods are assessed under het…
- arXiv cs.AIresearchMulti-Attractor GNNs: Set-Valued Expressivity Beyond Unique Equilibria
arXiv:2609.35274v1 Announce Type: cross Abstract: Recurrent and equilibrium graph neural networks (GNNs) often enforce a unique fixed point or use one training target per graph. Yet many combinatorial and scientific prob…
- arXiv cs.AIresearchReverse Sequential Proportional Approval Voting Rule: Proportionality and Approximation Guarantees
arXiv:2609.35331v1 Announce Type: cross Abstract: We study the Reverse Sequential Proportional Approval Voting Rule (RevSeqPAV) in approval-based committee elections. Despite its historical prominence and practical use,…
- arXiv cs.AIresearchLarge Language Models for Automated Cross-Domain Machine Learning Task Type Identification: A Benchmark Dataset and Evaluation
arXiv:2609.35335v1 Announce Type: cross Abstract: Machine learning task type identification is essential for constructing valid ML pipelines, yet in practice it is typically specified manually. We investigate whether lar…
- arXiv cs.AIresearchFrom Data to Program: Fast & Direct Generative Program Inference from Empirical Data
arXiv:2609.35348v1 Announce Type: cross Abstract: Estimating probability densities from a finite set of samples typically requires dataset-specific model fitting. We introduce PRODiGI, a pretrained data-to-program model…
- arXiv cs.AIresearchd-OPD: Future-Aware On-Policy Distillation for Block Diffusion Language Models
arXiv:2609.35362v1 Announce Type: cross Abstract: Large language models (LLMs) typically generate text autoregressively (AR), predicting one token at a time. Block diffusion language models (dLLMs) instead generate block…
- arXiv cs.AIresearchDo Temporal Link Predictors Need Learned Memory? A Smoothed-Count Baseline with a Handful of Parameters
arXiv:2609.35364v1 Announce Type: cross Abstract: Many temporal link predictors summarize past interactions through learned node representations. We examine whether simple counts of recurring interaction patterns can pro…
- arXiv cs.AIresearchFrom Pixel to Poses: Object-centric Tool Manipulation Learning from Human Demonstrations
arXiv:2609.35375v1 Announce Type: cross Abstract: Scaling up robotic manipulation is primarily bottlenecked by the scarcity of real-world robot data. While recent approaches leverage human video demonstrations to mitigat…
- arXiv cs.AIresearchThe Hidden Ratio in Adam: Stable Structure, Compression, and Sign Dynamics
arXiv:2609.35392v1 Announce Type: cross Abstract: Adam is the default optimizer for training modern deep neural networks, yet its adaptive behavior remains poorly understood due to the complex interaction between its fir…
- arXiv cs.AIresearchAwarenessBench: Assessing Cognitive Capabilities of Language Models
arXiv:2609.35409v1 Announce Type: cross Abstract: As language models (LMs) exhibit increasingly consciousness-like behaviors, evaluating their cognitive abilities becomes essential. We introduce AwarenessBench, the first…
- arXiv cs.AIresearchSpectral Super-Resolution using Spatial-Spectral Residual Operator Networks
arXiv:2609.35410v1 Announce Type: cross Abstract: Spectral super-resolution of multispectral satellite images can enable high temporal- and spatial-resolution hyperspectral satellite imagery at a modest cost, significant…
- arXiv cs.AIresearchGLAD: Global-Local Adaptive Detector for Robust Speech Deepfake Detection
arXiv:2609.35411v1 Announce Type: cross Abstract: Recent advances in AI-based speech synthesis have enabled highly realistic speech, increasing the importance of speech deepfake detection (SDD) in preventing misuse. Whil…
- arXiv cs.AIresearchSemantic Prefix Oracles for LLM Decoding: Contracts and Differential Validation
arXiv:2609.35425v1 Announce Type: cross Abstract: Constrained decoding can enforce regular or context-free output formats, but many program-generation failures are semantic: scope, typing, and declaration effects depend…
- arXiv cs.AIresearchFrontier Learning: Training LLM Reasoners at the Edge of Capability
arXiv:2609.35426v1 Announce Type: cross Abstract: Reinforcement Learning-based post-training of Large Language Models (LLM) has been successfully applied to improve their reasoning capabilities. Existing pipelines primar…
- arXiv cs.AIresearchReSPO: Reshaped Sequence Policy Optimization for Gradient Starvation in Off-Policy Learning
arXiv:2609.35433v1 Announce Type: cross Abstract: Reinforcement learning from verifiable rewards (RLVR) frequently reuses rollouts across multiple policy updates, increasing the mismatch between the current policy and th…
- arXiv cs.AIresearchRiccati State Space Models: Non-iterative Parallelization for Nonlinear Sequence Modeling
arXiv:2609.35441v1 Announce Type: cross Abstract: State space models (SSMs) achieve efficient sequence processing because their affine state updates are closed under composition and can therefore be evaluated with an ass…
- arXiv cs.AIresearchCLIMB: A Clinical Multimorbidity Benchmark for Diagnosing Co-occurring Conditions through Multiturn Conversations
arXiv:2609.35462v1 Announce Type: cross Abstract: Patients often have several co-occurring clinical conditions, and the findings needed to identify and disambiguate them emerge over the course of a consultation. Evaluati…
- arXiv cs.AIresearchRethinking Causal Action Tokenization with Conditional Annealing in Flow Matching
arXiv:2609.35469v1 Announce Type: cross Abstract: Autoregressive Vision-Language-Action (VLA) models offer a scalable path to robot learning, yet existing action tokenizers treat tokenization as a compression problem, pr…
- arXiv cs.AIresearchSpontaneous Context Restoration: How Language Models Recover from Corrupted Inputs
arXiv:2609.35475v1 Announce Type: cross Abstract: Language models sometimes produce correct outputs even when their inputs are corrupted by deletion, replacement, or misspelling. We study the internal processes accompany…
- arXiv cs.AIresearchAnalog Computing revisited: A fully analog and minimalistic Damage Detector for Ultrasonic Testing enabling Material-Integrated Structural Health Monitoring
arXiv:2609.35478v1 Announce Type: cross Abstract: Ultrasonic Testing (UT) is commonly used to detect damage in structures, e.g., metal plates. A sensor acquires Ultrasonic waves, e.g., by using PZT transducers. The time-…
- arXiv cs.AIresearchFrom Scores to Samples: Elastic Forcing for Autoregressive Video Generation
arXiv:2609.35491v3 Announce Type: replace-cross Abstract: Few-step autoregressive video generation commonly relies on Distribution Matching Distillation (DMD), requiring a bidirectional diffusion teacher and an online fa…
- arXiv cs.AIresearchSolveEdit: Benchmarking Visual Problem Solving in Generative Models
arXiv:2609.35504v1 Announce Type: cross Abstract: Machine intelligence is often evaluated through abstract reasoning problems, yet many real-world problems are visual, such as arranging objects, repairing layouts, or tra…
- arXiv cs.AIresearchImproving Generative Model Self-Training with Geometrically Modified Outputs
arXiv:2609.35512v1 Announce Type: cross Abstract: Self-training generative models - the continued improvement of a model using its own outputs - is becoming increasingly important as high-quality training data becomes sc…
- arXiv cs.AIresearchBeyond Token Scale: Chunk-Level Sparse Autoencoders for Reliable Semantic Feature Discovery
arXiv:2609.35521v1 Announce Type: cross Abstract: Sparse autoencoders (SAEs) expose features that help us understand and steer language models, but faithful reconstruction does not guarantee informative concepts. Token-l…
- arXiv cs.AIresearchLet the Neurons Die: Exploiting ReLU-Induced Model Degradation
arXiv:2609.35528v1 Announce Type: cross Abstract: Rectified linear unit (ReLU) networks can suffer from dying neurons, where units with persistently negative pre-activations produce zero outputs, blocking gradients throu…
- arXiv cs.AIresearchLess Sycophancy, Stronger Refusal? Lessons for AI Safety from Mechanistic Interpretability
arXiv:2609.35544v1 Announce Type: cross Abstract: Reliable refusal of harmful requests is essential to the safe deployment of language models. Because excessive eagerness to please users may undermine existing refusal ca…
- arXiv cs.AIresearchGraph World Models for Constrained Epidemic Policy Planning
arXiv:2609.35545v2 Announce Type: replace-cross Abstract: Epidemic policy planning often requires coordination between geographical regions, taking into account mobility-driven spillovers and how to make use of limited r…
- arXiv cs.AIresearchAlmieyar: A Culturally Grounded Benchmark for Multi-Dialect Arabic Speech Recognition
arXiv:2609.35564v1 Announce Type: cross Abstract: Arabic speech technology has largely focused on Modern Standard Arabic, leaving the living dialects spoken by hundreds of millions under-served. We introduce ALMIEYAR, a…
- arXiv cs.AIresearchF4R: Failure-Driven Recognition, Reconstruction, Refinement, and Redeployment for Continual Robot Self-Improvement
arXiv:2609.35575v2 Announce Type: replace-cross Abstract: The real-world performance of current vision-language-action models is fundamentally constrained by the limited coverage of expert demonstrations and their insuff…
- arXiv cs.AIresearchFactorEngram: Factorized N-gram Memory with Basis-Level Gating for Language Models
arXiv:2609.35578v1 Announce Type: cross Abstract: Lookup-based memory has been a promising way to scale the parameters of large language models (LLMs). It retrieves learned representations of local token patterns, such a…
- arXiv cs.AIresearchQC-Stark: A Multi-Task Benchmark Revealing Capability Dissociations in LLMs Evaluated on Quantum Computing Tasks
arXiv:2609.35581v1 Announce Type: cross Abstract: We introduce QC-Stark, a benchmark for evaluating large language models (LLMs) on 11 quantum computing (QC) tasks, spanning circuit construction, debugging, compilation,…
- arXiv cs.AIresearchTwist, Don't Tilt: Trajectory-Exact Constrained Decoding for Masked Diffusion Models
arXiv:2609.35609v1 Announce Type: cross Abstract: Constrained decoding for Masked Diffusion Language Models (MDLMs) aims to ensure that generated outputs satisfy a specified structure or syntax constraint. MDLMs generate…
- arXiv cs.AIresearchBehavioral Foundation Models for Quality Diversity
arXiv:2609.35615v1 Announce Type: cross Abstract: Behavioral Foundation Models (BFMs) are an emerging paradigm in reinforcement learning, playing a role analogous to large language models in natural language processing:…
- arXiv cs.AIresearchDR-net-Mamba: Selective State-Space Modeling for Long-Range ECG Time-Series Denoising
arXiv:2609.35634v1 Announce Type: cross Abstract: Electrocardiogram (ECG) recordings are corrupted by non-stationary noise sources that degrade diagnostic reliability, particularly in ambulatory and long-duration recordi…
- arXiv cs.AIresearchCMDO: A Cognitive Memory-Driven Optimization Algorithm for Adaptive Population-Based Search
arXiv:2609.35657v1 Announce Type: cross Abstract: Population-based optimization methods often use previous search information through successful solutions, parameter adaptation, or operator performance, but they rarely r…
- arXiv cs.AIresearchMS-GLA: Multi-Scale Gated Linear Attention for Addressing Representational Bottlenecks via Multi-Temporal Resolution
arXiv:2609.35664v2 Announce Type: replace-cross Abstract: Gated Linear Attention (GLA) Transformers advance linear recurrent models through data-dependent gating, but face a core limitation: the fixed-capacity memory mat…
- arXiv cs.AIresearchRethinking Circuit Evaluation: Do Circuits Explain Model Errors?
arXiv:2609.35686v1 Announce Type: cross Abstract: Mechanistic interpretability (MI) aims to explain a model's behaviour through analyzing its internal computations; circuit-based explanations aim to isolate these computa…
- arXiv cs.AIresearchDistillation Defenses Easily Break After Reinforcement Learning
arXiv:2609.35699v1 Announce Type: cross Abstract: Distillation attacks copy the reasoning capabilities of closed-source large language models, allowing bad actors to replicate state-of-the-art performance at low cost. At…
- arXiv cs.AIresearchA Unified Uncertainty Representation for Graph Neural Networks via Doubly-Spectral Stochastic Expansion
arXiv:2609.35703v1 Announce Type: cross Abstract: Reliable deployment of graph neural networks requires calibration, out-of-distribution (OOD) detection, and robustness to distribution shift, yet existing methods address…
- arXiv cs.AIresearchX-Reset: Scaling Object-Centric Reinforcement Learning via Cross-Embodiment Resets
arXiv:2609.35715v1 Announce Type: cross Abstract: Reinforcement learning (RL) in simulation can train dexterous manipulation policies without robot demonstrations, but training a single generalist policy with task-agnost…
- arXiv cs.AIresearchCopy the Same, Distill the Difference: Initializing Linear Vision Transformers
arXiv:2609.35745v1 Announce Type: cross Abstract: Linear Vision Transformers (ViTs) are designed to replace the attention in Softmax ViTs with the linear-complexity attention operator for more efficient token routing, bu…
- arXiv cs.AIresearchHow to Loop MoE: Flatten the Experts, Untie the Attention
arXiv:2609.35751v1 Announce Type: cross Abstract: Looped Transformers reuse one block of layers several times: by spending extra computation they push a model of fixed size further, and so use its parameters more fully;…
- arXiv cs.AIresearchLearning Native Reflection in Unified Models with Interleaved Reinforcement Learning
arXiv:2609.35767v2 Announce Type: replace-cross Abstract: Unified multimodal models can both look at and render images, so in principle they can repair their own generations: diagnose what an image gets wrong, revise it,…
- arXiv cs.AIresearchTelescopic Language Models
arXiv:2609.35769v1 Announce Type: cross Abstract: One deployed language model must often serve many compute budgets, yet serving each budget still means a separate training or compression run per point. We train a Telesc…
- arXiv cs.AIresearchFurE: Efficient Instance-Specific 3D Fur Reconstruction without Animal-Fur Datasets
arXiv:2609.35770v1 Announce Type: cross Abstract: Realistic and editable animal fur reconstruction from multi-view images is challenging due to fine-scale detail, self-occlusion and obfuscation, and, unlike human hair, t…
- arXiv cs.AIresearchCategorical Approach to Conflict Resolution:A Corrected Correspondence between Category Theory and the Graph Model for Conflict Resolution
arXiv:2306.13961v3 Announce Type: replace Abstract: This note is a substantially revised version of the author's earlier preprint (version~1, 2023), which proposed a ``Categorical Graph Model for Conflict Resolution'' (C…