AI Research News Feeds for September 15th, 2025

AI RESEARCH PAPERS & ACADEMIC SOURCES

• Data-driven approximation of transfer operators for mean-field stochastic differential equations
• A Smooth Computational Transition in Tensor PCA
• A Computable Measure of Suboptimality for Entropy-Regularised Variational Objectives
• Zero-inflation in the Multivariate Poisson Lognormal Family
• Can Generative Geospatial Diffusion Models Excel as Discriminative Geospatial Foundation Models?
• Talk2PC: Enhancing 3D Visual Grounding through LiDAR and Radar Point Clouds Fusion for Autonomous Driving
• Dynamic Motion Blending for Versatile Motion Editing
• GROVE: A Generalized Reward for Learning Open-Vocabulary Physical Skill
• MedM-VL: What Makes a Good Medical LVLM?
• Self-Rewarding Large Vision-Language Models for Optimizing Prompts in Text-to-Image Generation
• Earth Observation Foundation Model PhilEO: Pretraining on the MajorTOM and FastTOM Datasets
• Geometry and Perception Guided Gaussians for Multiview-consistent 3D Generation from a Single Image
• Integrative Variational Autoencoders for Generative Modeling of an Image Outcome with Multiple Input Images
• Out-Of-Distribution Detection for Audio-visual Generalized Zero-Shot Learning: A General Framework
• Efficient and Effective Adaptation of Multimodal Foundation Models in Sequential Recommendation
• Orientation Scores should be a Piece of Cake
• Taccel: Scaling Up Vision-based Tactile Robotics via High-performance GPU Simulation
• Uncovering Neuroimaging Biomarkers of Brain Tumor Surgery with AI-Driven Methods
• GAMMA: Generalizable Alignment via Multi-task and Manipulation-Augmented Training for AI-Generated Image Detection
• Robustness and Diagnostic Performance of Super-Resolution Fetal Brain MRI
• Mask Consistency Regularization in Object Removal
• MagicMirror: A Large-Scale Dataset and Benchmark for Fine-Grained Artifacts Assessment in Text-to-Image Generation
• Detecting Text Manipulation in Images using Vision Language Models
• Adversarial robustness through Lipschitz-Guided Stochastic Depth in Neural Networks
• A Stochastic Birth-and-Death Approach for Street Furniture Geolocation in Urban Environments
• Compute Only 16 Tokens in One Timestep: Accelerating Diffusion Transformers with Cluster-Driven Feature Caching
• GARD: Gamma-based Anatomical Restoration and Denoising for Retinal OCT
• Immunizing Images from Text to Image Editing via Adversarial Cross-Attention
• Efficient Learned Image Compression Through Knowledge Distillation
• Ordinality of Visible-Thermal Image Intensities for Intrinsic Image Decomposition
• Compressed Video Quality Enhancement: Classifying and Benchmarking over Standards
• InfGen: A Resolution-Agnostic Paradigm for Scalable Image Synthesis
• Chord: Chain of Rendering Decomposition for PBR Material Estimation from Generated Texture Images
• HHI-Assist: A Dataset and Benchmark of Human-Human Interaction in Physical Assistance Scenario
• Polarization Denoising and Demosaicking: Dataset and Baseline Method
• GC-VLN: Instruction as Graph Constraints for Training-free Vision-and-Language Navigation
• The Weighting Game: Evaluating Quality of Explainability Methods
• GeoDE: a Geographically Diverse Evaluation Dataset for Object Recognition
• LoFi: Vision-Aided Label Generator for Wi-Fi Localization and Tracking
• Afford-X: Generalizable and Slim Affordance Reasoning for Task-oriented Manipulation
• Privacy-Preserving Automated Rosacea Detection Based on Medically Inspired Region of Interest Selection
• Investigating the Impact of Various Loss Functions and Learnable Wiener Filter for Laparoscopic Image Desmoking
• SCoDA: Self-supervised Continual Domain Adaptation
• Segment Anything for Cell Tracking
• Online 3D Multi-Camera Perception through Robust 2D Tracking and Depth-based Late Aggregation
• Augment to Segment: Tackling Pixel-Level Imbalance in Wheat Disease and Pest Segmentation
• An HMM-based framework for identity-aware long-term multi-object tracking from sparse and uncertain identification: use case on long-term tracking in livestock
• ISTASTrack: Bridging ANN and SNN via ISTA Adapter for RGB-Event Tracking
• FLARE-SSM: Deep State Space Models with Influence-Balanced Loss for 72-Hour Solar Flare Prediction
• TUNI: Real-time RGB-T Semantic Segmentation with Unified Multi-Modal Feature Extraction and Cross-Modal Feature Fusion
• Few-Part-Shot Font Generation
• Efficient and Accurate Downfacing Visual Inertial Odometry
• Hierarchical MLANet: Multi-level Attention for 3D Face Reconstruction From Single Images
• LaV-CoT: Language-Aware Visual CoT with Multi-Aspect Reward Optimization for Real-World Multilingual VQA
• Color Me Correctly: Bridging Perceptual Color Spaces and Text Embeddings for Improved Diffusion Generation
• BEVTraj: Map-Free End-to-End Trajectory Prediction in Bird's-Eye View with Deformable Attention and Sparse Goal Proposals
• Leveraging Multi-View Weak Supervision for Occlusion-Aware Multi-Human Parsing
• A Lightweight Ensemble-Based Face Image Quality Assessment Method with Correlation-Aware Loss
• Grad-CL: Source Free Domain Adaptation with Gradient Guided Feature Disalignment
• Scalable Training for Vector-Quantized Networks with 100% Codebook Utilization
• LayerLock: Non-collapsing Representation Learning with Progressive Freezing
• On the Geometric Accuracy of Implicit and Primitive-based Representations Derived from View Rendering Constraints
• Atomic Fact Decomposition Helps Attributed Question Answering
• Are LLMs Better than Reported? Detecting Label Errors and Mitigating Their Effect on Model Performance
• A 2-step Framework for Automated Literary Translation Evaluation: Its Promises and Pitfalls
• FinMTEB: Finance Massive Text Embedding Benchmark
• Alignment-Augmented Speculative Decoding with Alignment Sampling and Conditional Verification
• Humans Hallucinate Too: Language Models Identify and Correct Subjective Annotation Errors With Label-in-a-Haystack Prompts
• NileChat: Towards Linguistically Diverse and Culturally Aware LLMs for Local Communities
• Faster and Better LLMs via Latency-Aware Test-Time Scaling
• Comparing Apples to Oranges: A Dataset & Analysis of LLM Humour Understanding from Traditional Puns to Topical Jokes
• Australian Supermarket Object Set (ASOS): A Benchmark Dataset of Physical Objects and 3D Models for Robotics and Computer Vision
• Decomposing Visual Classification: Assessing Tree-Based Reasoning in VLMs
• Images in Motion?: A First Look into Video Leakage in Collaborative Deep Learning
• Purge-Gate: Backpropagation-Free Test-Time Adaptation for Point Clouds Classification via Token Purging
• Fine-Grained Cross-View Localization via Local Feature Matching and Monocular Depth Priors
• Patch-based Automatic Rosacea Detection Using the ResNet Deep Learning Framework
• Discrimination by LLMs: Cross-lingual Bias Assessment and Mitigation in Decision-Making and Summarisation
• Pragmatic Frames Evoked by Gestures: A FrameNet Brasil Approach to Multimodality in Turn Organization
• Topic-Guided Reinforcement Learning with LLMs for Enhancing Multi-Document Summarization
• CMHG: A Dataset and Benchmark for Headline Generation of Minority Languages in China
• Multi-Intent Recognition in Dialogue Understanding: A Comparison Between Smaller Open-Source LLMs
• Linguistic trajectories of bipolar disorder on social media
• !MSA at BAREC Shared Task 2025: Ensembling Arabic Transformers for Readability Assessment
• Querying Climate Knowledge: Semantic Retrieval for Scientific Discovery
• Arabic Large Language Models for Medical Text Generation
• Scaling Arabic Medical Chatbots Using Synthetic Data: Enhancing Generative AI with Synthetic Patient Records
• Prominence-aware automatic speech recognition for conversational speech
• Towards Reliable and Interpretable Document Question Answering via VLMs
• Incongruent Positivity: When Miscalibrated Positivity Undermines Online Supportive Conversations
• Beyond Token Limits: Assessing Language Model Performance on Long Text Classification
• Dropping Experts, Recombining Neurons: Retraining-Free Pruning for Sparse Mixture-of-Experts LLMs
• Long Context Automated Essay Scoring with Language Models
• RefactorCoderQA: Benchmarking LLMs for Multi-Domain Coding Question Solutions in Cloud and Edge Deployment
• DeepDive: Advancing Deep Search Agents with Knowledge Graphs and Multi-Turn RL
• Whisper Has an Internal Word Aligner
• VARCO-VISION-2.0 Technical Report
• UIO-LLMs: Unbiased Incremental Optimization for Long-Context LLMs
• Direct Judgement Preference Optimization
• Optimal Multi-Task Learning at Regularization Horizon for Speech Translation Task
• BIBERT-Pipe on Biomedical Nested Named Entity Linking at BioASQ 2025
• Natural Language Translation of Formal Proofs through Informalization of Proof Steps and Recursive Summarization along Proof Structure
• A Role-Aware Multi-Agent Framework for Financial Education Question Answering with LLMs
• Benchmarking Vision-Language Models on Chinese Ancient Documents: From OCR to Knowledge Reasoning
• Task-Oriented Multimodal Token Transmission in Resource-Constrained Multiuser Networks
• Diffusion Buffer: Online Diffusion-based Speech Enhancement with Sub-Second Latency
• Attacking Attention of Foundation Models Disrupts Downstream Tasks
• Malware Classification Leveraging NLP & Machine Learning for Enhanced Accuracy
• Constructive Universal Approximation and Sure Convergence for Multi-Layer Neural Networks
• Constraint Guided Model Quantization of Neural Networks
• Bayesian Sheaf Neural Networks
• When and How Does CLIP Enable Domain and Compositional Generalization?
• Local-Cloud Inference Offloading for LLMs in Multi-Modal, Multi-Task, Multi-Dialogue Settings
• A Unified Framework for Diffusion Bridge Problems: Flow Matching and Schr\"{o}dinger Matching into One
• Space Group Informed Transformer for Crystalline Materials Generation
• Efficient transformer adaptation for analog in-memory computing via low-rank adapters
• MoPD: Mixture-of-Prompts Distillation for Vision-Language Models
• Soft Diamond Regularizers for Deep Learning
• Early Detection of Visual Impairments at Home Using a Smartphone Red-Eye Reflex Test
• An Information-Theoretic Framework for Credit Risk Modeling: Unifying Industry Practice with Statistical Theory for Fair and Interpretable Scorecards
• Off Policy Lyapunov Stability in Reinforcement Learning
• FetalSleepNet: A Transfer Learning Framework with Spectral Equalisation Domain Adaptation for Fetal Sleep Stage Classification
• Model-agnostic post-hoc explainability for recommender systems
• MCL-AD: Multimodal Collaboration Learning for Zero-Shot 3D Anomaly Detection
• Robot guide with multi-agent control and automatic scenario generation with LLM
• Why does your graph neural network fail on some graphs? Insights from exact generalisation error
• Characterizing the Efficiency of Distributed Training: A Power, Performance, and Thermal Perspective
• WhisTLE: Deeply Supervised, Text-Only Domain Adaptation for Pretrained Speech Recognition Transformers
• Is Adversarial Training with Compressed Datasets Effective?
• Unveiling Group-Specific Distributed Concept Drift: A Fairness Imperative in Federated Learning
• Uncertainty Modeling in Graph Neural Networks via Stochastic Differential Equations
• Federated Multi-Agent Reinforcement Learning for Privacy-Preserving and Energy-Aware Resource Management in 6G Edge Networks
• A Certifiable Machine Learning-Based Pipeline to Predict Fatigue Life of Aircraft Structures
• Prompt Injection Attacks on LLM Generated Reviews of Scientific Publications
• Property prediction for ionic liquids without prior structural knowledge using limited experimental data: A data-driven neural recommender system leveraging transfer learning
• Proof of AutoML: SDN based Secure Energy Trading with Blockchain in Disaster Case
• GraphCSVAE: Graph Categorical Structured Variational Autoencoder for Spatiotemporal Auditing of Physical Vulnerability Towards Sustainable Post-Disaster Risk Reduction
• ARMA Block: A CNN-Based Autoregressive and Moving Average Module for Long-Term Time Series Forecasting
• Physics-informed sensor coverage through structure preserving machine learning
• Flow Straight and Fast in Hilbert Space: Functional Rectified Flow
• Vendi Information Gain for Active Learning and its Application to Ecology
• Inpainting-Guided Policy Optimization for Diffusion Large Language Models
• Multipole Semantic Attention: A Fast Approximation of Softmax Attention for Pretraining
• Powering Job Search at Scale: LLM-Enhanced Query Understanding in Job Matching Systems
• DCHO: A Decomposition-Composition Framework for Predicting Higher-Order Brain Connectivity to Enhance Diverse Downstream Applications
• Improving MLLM Historical Record Extraction with Test-Time Image
• Hybrid Adaptive Conformal Offline Reinforcement Learning for Fair Population Health Management
• One Head, Many Models: Cross-Attention Routing for Cost-Aware LLM Selection
• Variational Neural Networks for Observable Thermodynamics (V-NOTS)
• LoFT: Parameter-Efficient Fine-Tuning for Long-tailed Semi-Supervised Learning in Open-World Scenarios
• SciML Agents: Write the Solver, Not the Solution
• DyKen-Hyena: Dynamic Kernel Generation via Cross-Modal Attention for Multimodal Intent Recognition
• Data-Driven Energy Estimation for Virtual Servers Using Combined System Metrics and Machine Learning
• Neural Scaling Laws for Deep Regression
• Uncertainty-Aware Tabular Prediction: Evaluating VBLL-Enhanced TabPFN in Safety-Critical Medical Data
• AdvI2I: Adversarial Image Attack on Image-to-Image Diffusion models
• Tokens, the oft-overlooked appetizer: Large language models, the distributional hypothesis, and meaning
• Can LLM Prompting Serve as a Proxy for Static Analysis in Vulnerability Detection
• A Novel Approach to Balance Convenience and Nutrition in Meals With Long-Term Group Recommendations and Reasoning on Multimodal Recipes and its Implementation in BEACON
• Auxiliary Discrminator Sequence Generative Adversarial Networks (ADSeqGAN) for Few Sample Molecule Generation
• Prompt Programming: A Platform for Dialogue-based Computational Problem Solving with Generative AI Models
• JARVIS-VLA: Post-Training Large-Scale Vision Language Models to Play Visual Games with Keyboards and Mouse
• A Framework for Testing and Adapting REST APIs as LLM Tools
• The Precautionary Principle and the Innovation Principle: Incompatible Guides for AI Innovation Governance?
• AReaL: A Large-Scale Asynchronous Reinforcement Learning System for Language Reasoning
• HiLight: A Hierarchical Reinforcement Learning Framework with Global Adversarial Guidance for Large-Scale Traffic Signal Control
• Atherosclerosis through Hierarchical Explainable Neural Network Analysis
• SignClip: Leveraging Mouthing Cues for Sign Language Translation by Multimodal Contrastive Fusion
• We Need a New Ethics for a World of AI Agents
• Generalizing Beyond Suboptimality: Offline Reinforcement Learning Learns Effective Scheduling through Random Data
• I-Segmenter: Integer-Only Vision Transformer for Efficient Semantic Segmentation
• GLAM: Geometry-Guided Local Alignment for Multi-View VLP in Mammography
• Data distribution impacts the performance and generalisability of contrastive learning-based foundation models of electrocardiograms
• Multimodal SAM-adapter for Semantic Segmentation
• Standards in the Preparation of Biomedical Research Metadata: A Bridge2AI Perspective
• Multi-Turn Human-LLM Interaction Through the Lens of a Two-Way Intelligibility Protocol
• QuantX: A Framework for Hardware-Aware Quantization of Generative AI Workloads
• Analyzing the Impact of Adversarial Examples on Explainable Machine Learning
• Slaves to the Law of Large Numbers: An Asymptotic Equipartition Property for Perplexity in Generative Language Models
• A Conflicts-free, Speed-lossless KAN-based Reinforcement Learning Decision System for Interactive Driving in Roundabouts
• A Survey on Group Fairness in Federated Learning: Challenges, Taxonomy of Solutions and Directions for Future Research
• Tackling One Health Risks: How Large Language Models are leveraged for Risk Negotiation and Consensus-building
• An Autoencoder and Vision Transformer-based Interpretability Analysis of the Differences in Automated Staging of Second and Third Molars
• WALL: A Web Application for Automated Quality Assurance using Large Language Models
• SmartCoder-R1: Towards Secure and Explainable Smart Contract Generation with Security-Aware Group Relative Policy Optimization
• Adaptive Token Merging for Efficient Transformer Semantic Communication at the Edge
• Securing LLM-Generated Embedded Firmware through AI Agent-Driven Validation and Patching
• Drone-Based Multispectral Imaging and Deep Learning for Timely Detection of Branched Broomrape in Tomato Farms
• Exploring Expert Specialization through Unsupervised Training in Sparse Mixture of Experts
• Reinforcement learning for spin torque oscillator tasks
• Multimodal Mathematical Reasoning Embedded in Aerial Vehicle Imagery: Benchmarking, Analysis, and Exploration
• Predictive Spike Timing Enables Distributed Shortest Path Computation in Spiking Neural Networks
• Established Psychometric vs. Ecologically Valid Questionnaires: Rethinking Psychological Assessments in Large Language Models
• Realism Control One-step Diffusion for Real-World Image Super-Resolution
• Population-Aligned Persona Generation for LLM-based Social Simulation
• BenchECG and xECG: a benchmark and baseline for ECG foundation models
• Benchmark of stylistic variation in LLM-generated texts
• SI-FACT: Mitigating Knowledge Conflict via Self-Improving Faithfulness-Aware Contrastive Tuning
• Openness in AI and downstream governance: A global value chain approach
• A Multimodal RAG Framework for Housing Damage Assessment: Collaborative Optimization of Image Encoding and Policy Vector Retrieval
• ALIGNS: Unlocking nomological networks in psychological measurement through a large language model
• DiTTO-LLM: Framework for Discovering Topic-based Technology Opportunities via Large Language Model
• MultimodalHugs: Enabling Sign Language Processing in Hugging Face
• MITS: A Large-Scale Multimodal Benchmark Dataset for Intelligent Traffic Surveillance
• Structure Matters: Brain Graph Augmentation via Learnable Edge Masking for Data-efficient Psychiatric Diagnosis
• D-CAT: Decoupled Cross-Attention Transfer between Sensor Modalities for Unimodal Inference
• A Co-Training Semi-Supervised Framework Using Faster R-CNN and YOLO Networks for Object Detection in Densely Packed Retail Images
• Meta-Learning Reinforcement Learning for Crypto-Return Prediction
• Revisiting Actor-Critic Methods in Discrete Action Off-Policy Reinforcement Learning
• SWE-Effi: Re-Evaluating Software AI Agent System Effectiveness Under Resource Constraints
• Vibe Check: Understanding the Effects of LLM-Based Conversational Agents' Personality and Alignment on User Perceptions in Goal-Oriented Tasks
• Emulating Public Opinion: A Proof-of-Concept of AI-Generated Synthetic Survey Responses for the Chilean Case
• From Hugging Face to GitHub: Tracing License Drift in the Open-Source AI Ecosystem
• Automated Tuning for Diffusion Inverse Problem Solvers without Generative Prior Retraining
• Mutual Information Tracks Policy Coherence in Reinforcement Learning
• Forecasting Clicks in Digital Advertising: Multimodal Inputs and Interpretable Outputs
• Text-to-SQL Oriented to the Process Mining Domain: A PT-EN Dataset for Query Translation
• TalkPlayData 2: An Agentic Synthetic Data Pipeline for Multimodal Conversational Music Recommendation
• GeoGPT.RAG Technical Report
• AI-Powered Assistant for Long-Term Access to RHIC Knowledge
• Structured Information Matters: Explainable ICD Coding with Patient-Level Knowledge Graphs
• Cross-Layer Attention Probing for Fine-Grained Hallucination Detection
• Creativity Benchmark: A benchmark for marketing creativity for LLM models
• CTCC: A Robust and Stealthy Fingerprinting Framework for Large Language Models via Cross-Turn Contextual Correlation Backdoor
• The Non-Determinism of Small LLMs: Evidence of Low Answer Consistency in Repetition Trials of Standard Multiple-Choice Benchmarks
• Differential Robustness in Transformer Language Models: Empirical Evaluation Under Adversarial Text Attacks
• LLM-Based Instance-Driven Heuristic Bias In the Context of a Biased Random Key Genetic Algorithm
• Beyond I'm Sorry, I Can't: Dissecting Large Language Model Refusal
• Assisting Research Proposal Writing with Large Language Models: Evaluation and Refinement
• Generating Individual Travel Diaries Using Large Language Models Informed by Census and Land-Use Data
• Psychiatry-Bench: A Multi-Task Benchmark for LLMs in Psychiatry
• The Thinking Therapist: Training Large Language Models to Deliver Acceptance and Commitment Therapy using Supervised Fine-Tuning and Odds Ratio Policy Optimization
• HANRAG: Heuristic Accurate Noise-resistant Retrieval-Augmented Generation for Multi-hop Question Answering
• Human-AI Collaboration Increases Efficiency in Regulatory Writing
• How well can LLMs provide planning feedback in grounded environments?
• A Modular and Multimodal Generative AI Framework for Urban Building Energy Data: Generating Synthetic Homes
• Towards an AI-based knowledge assistant for goat farmers based on Retrieval-Augmented Generation
• LLMs as Agentic Cooperative Players in Multiplayer UNO
• The (R)evolution of Scientific Workflows in the Agentic AI Era: Towards Autonomous Science
• Evaluation of Black-Box XAI Approaches for Predictors of Values of Boolean Formulae
• GAMA: A General Anonymizing Multi-Agent System for Privacy Preservation Enhanced by Domain Rules and Disproof Method
• XAgents: A Unified Framework for Multi-Agent Cooperation via IF-THEN Rules and Multipolar Task Processing Graph
• AI Harmonics: a human-centric and harms severity-adaptive AI risk assessment framework
• Online Robust Planning under Model Uncertainty: A Sample-Based Approach
• The Morality of Probability: How Implicit Moral Biases in LLMs May Shape the Future of Human-AI Symbiosis
• Abduct, Act, Predict: Scaffolding Causal Inference for Automated Failure Attribution in Multi-Agent Systems