AI news archive.
10,661 articles filed under Research, newest first, from the outlets listed on the sources page.
- arXiv cs.AIresearchBandits with Multiple Optimal Arms: Minimax Regret and Non-Adaptivity
arXiv:2609.38659v2 Announce Type: replace-cross Abstract: We study multi-armed bandits (MAB) with multiple optimal arms, motivated by the fact that many practical decision making problems admit multiple correct answers.…
- arXiv cs.AIresearchUnderstanding Off- vs On-Policy Distillation: A Tale of Distinct Training Objectives
arXiv:2609.38666v1 Announce Type: cross Abstract: On-policy distillation (OPD) learns from teacher feedback on student-generated responses and has shown promise in reducing forgetting relative to supervised fine-tuning (…
- arXiv cs.AIresearchProvable Test-Time Scaling for Beam Search in LLM Reasoning
arXiv:2609.38672v1 Announce Type: cross Abstract: Beam-search-based test-time methods provide an effective way to improve large language model (LLM) performance on long-horizon generation by pruning invalid reasoning pat…
- arXiv cs.AIresearchReGain: Restoring Subject Fidelity in Personalization on Synthetic Images
arXiv:2609.38680v1 Announce Type: cross Abstract: Text-to-image diffusion models are personalized to a subject by DreamBooth fine-tuning on a handful of its images. Increasingly, these images come from a diffusion model…
- arXiv cs.AIresearchUnveiling the Value of Motion for Cinematic Camera Trajectories
arXiv:2609.38683v1 Announce Type: cross Abstract: Cinematic camera motion is a fundamental storytelling tool, defined not only by where the camera is positioned in the scene, but also by how it moves in terms of directio…
- arXiv cs.AIresearchAlways-On Experimentation
arXiv:2609.38695v1 Announce Type: cross Abstract: Generative AI has dramatically accelerated the rate at which new treatments---from novel pharmaceuticals to online marketing campaigns---can be conceived and deployed. As…
- arXiv cs.AIresearchCascadia: A Control-Plane-Free Alternative to Hyperconverged AI Infrastructure
arXiv:2609.38697v1 Announce Type: cross Abstract: We present Cascadia, a system for serving large language models on fleets of commodity Intel AIPCs using their CPU, integrated-GPU, and NPU resources. Every node embeds i…
- arXiv cs.AIresearchSpatialCORE: Confidence-Aware Grounded Spatial Reasoning in Large Vision--Language Models
arXiv:2609.38716v1 Announce Type: cross Abstract: Large Vision-Language Models (LVLMs) have made remarkable progress across visual perception tasks, yet spatial reasoning remains a persistent weakness, especially for que…
- arXiv cs.AIresearchSoft Spatial Reasoning
arXiv:2609.38717v1 Announce Type: cross Abstract: Large Vision-Language Models (LVLMs) commonly perform spatial reasoning through chain-of-thought (CoT), encoding intermediate reasoning as autoregressive sequences of dis…
- arXiv cs.AIresearchAdaptive-GEPA: Make Your Harness Fit Heterogeneous Requests
arXiv:2609.38762v1 Announce Type: cross Abstract: Reflective optimizers such as GEPA improve language model prompts from execution traces and evaluator feedback; full-program extensions can also rewrite tools and control…
- arXiv cs.AIresearchdattri-LLM: A Unified and Efficient Library for Training Data Attribution at LLM Scale
arXiv:2609.38767v2 Announce Type: replace-cross Abstract: Training data attribution (TDA) estimates the contribution of individual training examples to model outputs. Most scalable TDA methods rely on per-example gradien…
- arXiv cs.AIresearchLearning Under Forgetting: Statistical Support-Selective Retention in Stochastic Training Dynamics
arXiv:2609.38768v1 Announce Type: cross Abstract: Prior work has shown that neural networks exhibit implicit biases toward low-complexity structure (e.g., spectral bias), memorization dynamics, and compression-like effec…
- arXiv cs.AIresearchDistilling Diffusion Score Discrepancy for Efficient Training Data Attribution
arXiv:2609.38776v2 Announce Type: replace-cross Abstract: Training data attribution for diffusion models aims to identify the training samples that influence a generated instance, but existing methods either require cost…
- arXiv cs.AIresearchRAST: Resolution-Aware Privileged Structure Transfer for Low-Resolution Audio Activity Recognition
arXiv:2609.38780v1 Announce Type: cross Abstract: Audio is increasingly used for human activity recognition (HAR) because it captures object interactions, environmental events, and contextual cues in everyday environment…
- arXiv cs.AIresearchChartDensity-Bench: Benchmarking MLLMs for Numerical Data Reconstruction under Visual Density
arXiv:2609.38781v1 Announce Type: cross Abstract: Multimodal large language models (MLLMs) offer a promising approach for recovering numerical data from scientific charts, but their ability to reconstruct chart data from…
- arXiv cs.AIresearchUncovering Uncontrolled Repetition through Residual Stream Dynamics
arXiv:2609.38802v1 Announce Type: cross Abstract: Uncontrolled repetition can prolong autoregressive generation in large language models (LLMs) and enable resource consumption attacks. Prior analyses of repetitive genera…
- arXiv cs.AIresearchCRAFT: Causal Responsibility and Failure Tracing in Medical Vision Language Models
arXiv:2609.38810v3 Announce Type: replace-cross Abstract: As vision language models are increasingly deployed in clinical diagnosis, understanding how they internally resolve competing visual and textual signals becomes…
- arXiv cs.AIresearchFuture Video Generation Better Aligns with the Human Visual Cortex than Observed Video
arXiv:2609.38819v1 Announce Type: cross Abstract: Studying the alignment between the internal representations of vision models and the responses of the visual cortex to the same observed visual stimuli has enabled us to…
- arXiv cs.AIresearchBARRAC: Adaptation of an English Aspect-based Sentiment Analysis Approach for Classification Tasks in Arabic Dialects
arXiv:2609.38820v1 Announce Type: cross Abstract: With the rapid growth of Arabic NLP, several models, datasets and benchmarks have been reported. This paper asks whether approaches developed for majority languages like…
- arXiv cs.AIresearchscTrilemma: Balancing Identity, Invariance, and Fidelity in Single-Cell Representation Learning
arXiv:2609.38840v2 Announce Type: replace-cross Abstract: Single-cell RNA-seq representation learning is fundamentally label-free: cell identities, states, and contexts are not fixed training targets, so what constitutes…
- arXiv cs.AIresearchScoring Higher, Answering Worse: Mitigating Reward Hacking in Rubric-Based RL via Protocol-Level Rubrics
arXiv:2609.38847v1 Announce Type: cross Abstract: Rubric-based reinforcement learning (Rubric-RL) trains language models where no verifier exists. A judge checks each criterion of a rubric, and the verdicts are aggregate…
- arXiv cs.AIresearchWhere MLLMs Fail and Why: Causal Task Decomposition for Capability Failure Diagnosis
arXiv:2609.38851v1 Announce Type: cross Abstract: End-to-end accuracy on compositional tasks records how often MLLMs fail, but cannot distinguish whether a failure reflects an intrinsic deficit in the targeted capability…
- arXiv cs.AIresearchMitigating the Length-Scaling Tax with Online Distillation
arXiv:2609.38854v1 Announce Type: cross Abstract: Length scaling during reinforcement-learning (RL) post-training is often viewed as a sign of improved reasoning ability, especially on difficult problems, but may also ma…
- arXiv cs.AIresearchOnline Evolution Strategy for Flow-Matching VLA Policies via Self-Supervised Trajectory Distribution Optimization
arXiv:2609.38855v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models based on generative frameworks, such as Flow Matching, have recently achieved impressive performance in robotic manipulation. Unlike d…
- arXiv cs.AIresearchOptimal Design for Active Preference Learning with Biased LLM Judges
arXiv:2609.38860v1 Announce Type: cross Abstract: Learning from human preferences is central to large language model (LLM) alignment, but human preference annotation is costly. Active preference learning reduces this cos…
- arXiv cs.AIresearchGeoNest: Learning to Select Failure-Aware Neighborhoods for the Irregular Knapsack Problem in a Circular Container
arXiv:2609.38863v1 Announce Type: cross Abstract: The two-dimensional irregular knapsack problem in a fixed circular container is an important combinatorial optimization problem for maximizing material utilization in man…
- arXiv cs.AIresearchAudio Token Attention Is Predictable Before the Language Model Runs
arXiv:2609.38878v1 Announce Type: cross Abstract: A large audio language model (LALM) turns a minute of speech into 750-1,500 tokens and prefills every one. Image-token pruning often cuts after the language model's first…
- arXiv cs.AIresearchDoes Learning Protein Folding Generalize to Broader Reasoning?
arXiv:2609.38879v1 Announce Type: cross Abstract: Large language models rely heavily on human text, which often conveys surface answers rather than the spatial and structural logic behind them. Protein folding is a natur…
- arXiv cs.AIresearchRight Answers, Costly Models: The Efficiency Gap in LLM-based Optimization Modeling
arXiv:2609.38884v1 Announce Type: cross Abstract: Optimization modeling formulates real-world decision problems as mathematical programs that solvers can use to find optimal decisions. Large language models (LLMs) can au…
- arXiv cs.AIresearchLearning Continuous Neural Representation of Stochastic Hybrid Systems
arXiv:2609.38893v1 Announce Type: cross Abstract: A stochastic hybrid system (SHS) is governed by a stochastic differential equation (SDE) describing the continuous dynamics and a Markov reset kernel triggered on the gua…
- arXiv cs.AIresearchUnmerge: Efficient Machine Unlearning via Task Arithmetic
arXiv:2609.38895v1 Announce Type: cross Abstract: Approximate machine unlearning seeks to remove the influence of a forget set from a trained model without full retraining. Existing gradient-based methods require data-de…
- arXiv cs.AIresearchFFASR: Benchmarking Far-Field Automatic Speech Recognition using High-Fidelity Simulated RIRs
arXiv:2609.38897v1 Announce Type: cross Abstract: Far-field automatic speech recognition(ASR) degrades under reverberation, noise, and talker motion, yet the benchmarks that drive model selection emphasize close-micropho…
- arXiv cs.AIresearchSceneJail: Exploiting Video Scenario Context to Jailbreak Multimodal LLMs
arXiv:2609.38899v1 Announce Type: cross Abstract: Video Multimodal Large Language Models (Video-MLLMs) support reasoning over video inputs, yet remain vulnerable to jailbreak attacks that elicit policy-violating response…
- arXiv cs.AIresearchDAMPER: Return-Prioritized Gradient Control for Smooth Policies
arXiv:2609.38903v1 Announce Type: cross Abstract: Actor-critic methods achieve strong performance in continuous control, but their policies can produce highly oscillatory actions. A common remedy is to add auxiliary smoo…
- arXiv cs.AIresearchCharacterizing Questioning Patterns and Student Engagement Through Contextual Analysis of Real-Time Classroom Interactions
arXiv:2609.38907v1 Announce Type: cross Abstract: Real-time classroom polling is now routine, yet the data it produces is usually read narrowly, as a correctness score or a headcount. Such readings say little about what…
- arXiv cs.AIresearchCellMSA: Context Modeling for Single-Cell Representation Learning
arXiv:2609.38908v1 Announce Type: cross Abstract: Single-cell transcriptomics enables profiling of cellular states at unprecedented resolution, but its high dimensionality, sparsity, and technical batch effects pose sign…
- arXiv cs.AIresearchUnlearning Deceptive Behaviors in LLMs with Contrastive Forget Sets
arXiv:2609.38909v1 Announce Type: cross Abstract: Large language models often know the truth and say otherwise: a model that answers correctly when asked neutrally will affirm a user's mistaken belief, or misstate a fact…
- arXiv cs.AIresearchOn the Relaxation of Conditional Independence Assumption for Image Segmentation
arXiv:2609.38930v2 Announce Type: replace-cross Abstract: In semantic segmentation, a recent line of RankSEG methods directly optimizes Dice/IoU scores at inference time, improving alignment with evaluation metrics witho…
- arXiv cs.AIresearchAdaptive Self-Consistency: From Black-Box Sampling to Distribution-Valued Feedback
arXiv:2609.38931v1 Announce Type: cross Abstract: Self-consistency samples many reasoning trajectories and aggregates their final answers, treating the LLM as a black box that returns one answer per trajectory. Yet the f…
- arXiv cs.AIresearchSignal-Routed Temperature Scaling: Low-Capacity Risk-Conditioned Calibration for Small Validation Budgets
arXiv:2609.38936v1 Announce Type: cross Abstract: When a classifier is recalibrated from only a few thousand held-out examples, the capacity of the calibration map becomes a statistical design choice rather than a purely…
- arXiv cs.AIresearchDrivingBench: Can Vision-Language Models Drive a Toyota Corolla?
arXiv:2609.38948v1 Announce Type: cross Abstract: Frontier models excel at many digital benchmarks, yet their ability to drive a real car, an everyday human skill, remains largely untested. We present DrivingBench, to ou…
- arXiv cs.AIresearchLoop-Free Inverse Reinforcement Learning via Sequential Value Recovery with Q-Score Matching
arXiv:2609.38955v1 Announce Type: cross Abstract: Inverse Reinforcement Learning (IRL) aims to recover a reward function that explains expert demonstrations. Existing IRL methods typically rely on a bi-level optimization…
- arXiv cs.AIresearchMaking LLMs Say What They Think: Measuring and Improving CoT-Interpretability Alignment
arXiv:2609.38972v2 Announce Type: replace-cross Abstract: Chain-of-thought (CoT) traces often serve as a proxy for how Large Language Models (LLMs) arrive at their answers. However, growing evidence shows that models' Co…
- arXiv cs.AIresearchScale-Split Neural Operator for Memory- and Data-Efficient 3D Turbulence Prediction
arXiv:2609.38977v1 Announce Type: cross Abstract: Neural surrogates have emerged as fast alternatives to the numerical simulation of three-dimensional turbulence. However, training them at high resolution remains challen…
- arXiv cs.AIresearchSparse-WAM: Accelerating World Action Models via Action-Guided Sparse Imagination
arXiv:2609.38984v1 Announce Type: cross Abstract: World-action models (WAMs) leverage pretrained video models to improve generalization in robot control by jointly predicting future visual states and actions. This capabi…
- arXiv cs.AIresearchThe Invisible Language Tax: Token Premiums of French and Regional Languages in 2026 LLM Tokenizers, and a French-Optimized Prototype
arXiv:2609.39001v1 Announce Type: cross Abstract: LLM services are billed per token and context windows are measured in tokens, yet the number of tokens needed for the same content varies across languages. We measure thi…
- arXiv cs.AIresearchAn Uncertainty-Guided Digital Twin Framework for Online Adaptive Proton Therapy in Head and Neck Cancer: A Feasibility Study
arXiv:2609.39010v1 Announce Type: cross Abstract: Objective: Head and neck (HN) proton therapy spans six to seven weeks of anatomical change, while offline replanning takes about a week. We present an uncertainty-guided…
- arXiv cs.AIresearchTED:Text-Axis Evidence Decomposition for Prompted Anomaly Localization
arXiv:2609.39033v1 Announce Type: cross Abstract: CLIP is a powerful vision-language model, but it was not designed for fine-grained defect localization; CLIP-based anomaly detectors therefore adapt it with prompts or li…
- arXiv cs.AIresearchSwitching Linear Attention
arXiv:2609.39034v1 Announce Type: cross Abstract: Designing expressive sequence layers with efficient inference remains a central challenge in modern machine learning. Standard softmax attention achieves excellent sequen…
- arXiv cs.AIresearchCycle-Aware Autoencoder with Cross-SignalConsistency for Railway Door Anomaly Detection
arXiv:2609.39035v1 Announce Type: cross Abstract: Passenger access doors are safety-critical subsystems in railway vehicles, yet detecting abnormal door behavior in real operation is challenging because faults are rare,…