AI Research News Feeds for December 25th, 2025

AI RESEARCH PAPERS & ACADEMIC SOURCES

• Steering Vision-Language Pre-trained Models for Incremental Face Presentation Attack Detection
• Decoding Predictive Inference in Visual Language Processing via Spatiotemporal Neural Coherence
• Transductive Visual Programming: Evolving Tool Libraries from Experience for Spatial Reasoning
• Beyond Context: Large Language Models Failure to Grasp Users Intent
• ReaSeq: Unleashing World Knowledge via Reasoning for Sequential Modeling
• Improving Neural Question Generation using World Knowledge
• Sequence to Sequence Reward Modeling: Improving RLHF by Language Feedback
• VL4Gaze: Unleashing Vision-Language Models for Gaze Following
• OccuFly: A 3D Vision Benchmark for Semantic Scene Completion from the Aerial Perspective
• Learning to Sense for Driving: Joint Optics-Sensor-Model Co-Design for Semantic Segmentation
• Input-Adaptive Visual Preprocessing for Efficient Fast Vision-Language Model Inference
• ALIVE: An Avatar-Lecture Interactive Video Engine with Content-Aware Retrieval for Real-Time Interaction
• Lightweight framework for underground pipeline recognition and spatial localization based on multi-view 2D GPR images
• NeRV360: Neural Representation for 360-Degree Videos with a Viewport Decoder
• Beyond Weight Adaptation: Feature-Space Domain Injection for Cross-Modal Ship Re-Identification
• DGSAN: Dual-Graph Spatiotemporal Attention Network for Pulmonary Nodule Malignancy Prediction
• Benchmarking and Enhancing VLM for Compressed Image Understanding
• PanoGrounder: Bridging 2D and 3D with Panoramic Scene Representations for VLM-based 3D Visual Grounding
• Self-supervised Multiplex Consensus Mamba for General Image Fusion
• Quantile Rendering: Efficiently Embedding High-dimensional Feature on 3D Gaussian Splatting
• Reasoning-Driven Amodal Completion: Collaborative Agents and Perceptual Evaluation
• Beyond Artifacts: Real-Centric Envelope Modeling for Reliable AI-Generated Image Detection
• SPOT!: Map-Guided LLM Agent for Unsupervised Multi-CCTV Dynamic Object Tracking
• XGrid-Mapping: Explicit Implicit Hybrid Grid Submaps for Efficient Incremental Neural LiDAR Mapping
• X-ray Insights Unleashed: Pioneering the Enhancement of Multi-Label Long-Tail Data
• PUFM++: Point Cloud Upsampling via Enhanced Flow Matching
• MVInverse: Feed-forward Multi-view Inverse Rendering in Seconds
• Learning from Next-Frame Prediction: Autoregressive Video Modeling Encodes Effective Representations
• Granular-ball Guided Masking: Structure-aware Data Augmentation
• FluencyVE: Marrying Temporal-Aware Mamba with Bypass Attention for Video Editing
• Efficient and Robust Video Defense Framework against 3D-field Personalized Talking Face
• Multi-Attribute guided Thermal Face Image Translation based on Latent Diffusion Model
• Next-Scale Prediction: A Self-Supervised Approach for Real-World Image Denoising
• A Large-Depth-Range Layer-Based Hologram Dataset for Machine Learning-Based 3D Computer-Generated Holography
• Matrix Completion Via Reweighted Logarithmic Norm Minimization
• Optical Flow-Guided 6DoF Object Pose Tracking with an Event Camera
• Beyond Pixel Simulation: Pathology Image Generation via Diagnostic Semantic Tokens and Prototype Control
• Multimodal Skeleton-Based Action Representation Learning via Decomposition and Composition
• UniPR-3D: Towards Universal Visual Place Recognition with Visual Geometry Grounded Transformer
• T2AV-Compass: Towards Unified Evaluation for Text-to-Audio-Video Generation
• UniRec-0.1B: Unified Text and Formula Recognition with 0.1B Parameters
• FreeInpaint: Tuning-free Prompt Alignment and Visual Rationality Enhancement in Image Inpainting
• MarineEval: Assessing the Marine Intelligence of Vision-Language Models
• TGC-Net: A Structure-Aware and Semantically-Aligned Framework for Text-Guided Medical Image Segmentation
• ORCA: Object Recognition and Comprehension for Archiving Marine Species
• A Turn Toward Better Alignment: Few-Shot Generative Adaptation with Equivariant Feature Rotation
• Towards Arbitrary Motion Completing via Hierarchical Continuous Representation
• UltraShape 1.0: High-Fidelity 3D Shape Generation via Scalable Geometric Refinement
• VisRes Bench: On Evaluating the Visual Reasoning Capabilities of VLMs
• Human Motion Estimation with Everyday Wearables
• Latent Implicit Visual Reasoning
• Leveraging Lightweight Entity Extraction for Scalable Event-Based Image Retrieval
• SegMo: Segment-aligned Text to 3D Human Motion Generation
• DreaMontage: Arbitrary Frame-Guided One-Shot Video Generation
• AnyAD: Unified Any-Modality Anomaly Detection in Incomplete Multi-Sequence MRI
• ACD: Direct Conditional Control for Video Diffusion Models via Attention Supervision
• GriDiT: Factorized Grid-Based Diffusion for Efficient Long Image Sequence Generation
• Surgical Scene Segmentation using a Spike-Driven Video Transformer with Real-Time Potential
• Post-Processing Mask-Based Table Segmentation for Structural Coordinate Extraction
• AndroidLens: Long-latency Evaluation with Nested Sub-targets for Android GUI Agents
• TICON: A Slide-Level Tile Contextualizer for Histopathology Representation Learning
• Fast SAM2 with Text-Driven Token Pruning
• Streaming Video Instruction Tuning
• Beyond Memorization: A Multi-Modal Ordinal Regression Benchmark to Expose Popularity Bias in Vision-Language Models
• HiStream: Efficient High-Resolution Video Generation via Redundancy-Eliminated Streaming
• Flow Gym
• Language-Guided Grasp Detection with Coarse-to-Fine Learning for Robotic Manipulation
• TexAvatars : Hybrid Texel-3D Representations for Stable Rigging of Photorealistic Gaussian Head Avatars
• Equivariant Multiscale Learned Invertible Reconstruction for Cone Beam CT: From Simulated to Real Data
• Schr\"odinger's Navigator: Imagining an Ensemble of Futures for Zero-Shot Object Navigation
• RoboSafe: Safeguarding Embodied Agents via Executable Safety Logic
• Automatic Replication of LLM Mistakes in Medical Conversations
• Learning from Neighbors with PHIBP: Predicting Infectious Disease Dynamics in Data-Sparse Environments
• Enhancing diffusion models with Gaussianization preprocessing
• Towards Better Search with Domain-Aware Text Embeddings for C2C Marketplaces
• Critical Points of Degenerate Metrics on Algebraic Varieties: A Tale of Overparametrization
• Agentic Multi-Persona Framework for Evidence-Aware Fake News Detection
• zkFL-Health: Blockchain-Enabled Zero-Knowledge Federated Learning for Medical AI Privacy
• DexAvatar: 3D Sign Language Reconstruction with Hand and Body Pose Priors
• Blurb-Refined Inference from Crowdsourced Book Reviews using Hierarchical Genre Mining with Dual-Path Graph Convolutions
• LLM Personas as a Substitute for Field Experiments in Method Benchmarking
• Dyna-Style Reinforcement Learning Modeling and Control of Non-linear Dynamics
• Hierarchical Modeling Approach to Fast and Accurate Table Recognition
• Semantic Refinement with LLMs for Graph Representations
• Semi-Supervised Learning for Large Language Models Safety and Content Moderation
• AutoBaxBuilder: Bootstrapping Code Security Benchmarking
• ElfCore: A 28nm Neural Processor Enabling Dynamic Structured Sparse Training and Online Self-Supervised Learning with Activity-Dependent Weight Update
• A Community-Enhanced Graph Representation Model for Link Prediction
• Causal-driven attribution (CDA): Estimating channel influence without user-level data
• Assessing the Software Security Comprehension of Large Language Models
• LookPlanGraph: Embodied Instruction Following Method with VLM Graph Augmentation
• Variationally correct operator learning: Reduced basis neural operator with a posteriori error estimation
• Parallel Token Prediction for Language Models
• Autonomous Uncertainty Quantification for Computational Point-of-care Sensors
• Optimizing Decoding Paths in Masked Diffusion Models by Quantifying Uncertainty
• Explicit Group Sparse Projection with Applications to Deep Learning and NMF
• DATTA: Domain Diversity Aware Test-Time Adaptation for Dynamic Domain Shift Data Streams
• Optimal Control with Natural Images: Efficient Reinforcement Learning using Overcomplete Sparse Codes
• Predicting Metabolic Dysfunction-Associated Steatotic Liver Disease using Machine Learning Methods
• Deep Kronecker Network
• Eliciting Risk Aversion with Inverse Reinforcement Learning via Interactive Questioning
• Agnostic Process Tomography
• MatchMiner-AI: An Open-Source Solution for Cancer Clinical Trial Matching
• SA-DiffuSeq: Addressing Computational and Scalability Challenges in Long-Document Generation with Sparse Attention
• Adversarial Training for Failure-Sensitive User Simulation in Mental Health Dialogue Optimization
• Large Language Models Approach Expert Pedagogical Quality in Math Tutoring but Differ in Instructional and Linguistic Profiles
• Investigating Model Editing for Unlearning in Large Language Models
• Measuring Mechanistic Independence: Can Bias Be Removed Without Erasing Demographics?
• Semantic Deception: When Reasoning Models Can't Compute an Addition
• EssayCBM: Rubric-Aligned Concept Bottleneck Models for Transparent Essay Grading
• MediEval: A Unified Medical Benchmark for Patient-Contextual and Knowledge-Grounded Reasoning in LLMs
• How important is Recall for Measuring Retrieval Quality?
• Where Did This Sentence Come From? Tracing Provenance in LLM Reasoning Distillation
• Foundation Model-based Evaluation of Neuropsychiatric Disorders: A Lifespan-Inclusive, Multi-Modal, and Multi-Lingual Study
• Neural Probe-Based Hallucination Detection for Large Language Models
• Reflection Pretraining Enables Token-Level Self-Correction in Biological Sequence Models
• Distilling the Essence: Efficient Reasoning Distillation via Sequence Truncation
• Rethinking Supervised Fine-Tuning: Emphasizing Key Answer Tokens for Improved LLM Accuracy
• ClarifyMT-Bench: Benchmarking and Improving Multi-Turn Clarification for Conversational Large Language Models
• SpidR-Adapt: A Universal Speech Representation Model for Few-Shot Adaptation
• SMART SLM: Structured Memory and Reasoning Transformer, A Small Language Model for Accurate Document Assistance
• Your Reasoning Benchmark May Not Test Reasoning: Revealing Perception Bottleneck in Abstract Reasoning Benchmarks
• C2LLM Technical Report: A New Frontier in Code Retrieval via Adaptive Cross-Attention Pooling
• MegaRAG: Multimodal Knowledge Graph-Based Retrieval Augmented Generation
• Automated Red-Teaming Framework for Large Language Model Security Assessment: A Comprehensive Attack Generation and Detection System
• Zero-Training Temporal Drift Detection for Transformer Sentiment Models: A Comprehensive Analysis on Authentic Social Media Streams
• Enhancing Lung Cancer Treatment Outcome Prediction through Semantic Feature Engineering Using Large Language Models
• Real Time Detection and Quantitative Analysis of Spurious Forgetting in Continual Learning
• SHRP: Specialized Head Routing and Pruning for Efficient Encoder Compression
• Data-Free Pruning of Self-Attention Layers in LLMs
• Forecasting N-Body Dynamics: A Comparative Study of Neural Ordinary Differential Equations and Universal Differential Equations
• Q-RUN: Quantum-Inspired Data Re-uploading Networks
• MaskOpt: A Large-Scale Mask Optimization Dataset to Advance AI in Integrated Circuit Manufacturing
• Managing the Stochastic: Foundations of Learning in Neuro-Symbolic Systems for Software Engineering
• Dominating vs. Dominated: Generative Collapse in Diffusion Models
• Forward Only Learning for Orthogonal Neural Networks of any Depth
• Improving Cardiac Risk Prediction Using Data Generation Techniques
• Disentangling Fact from Sentiment: A Dynamic Conflict-Consensus Framework for Multimodal Fake News Detection
• HyDRA: Hierarchical and Dynamic Rank Adaptation for Mobile Vision Language Model
• Revisiting the Learning Objectives of Vision-Language Reward Models
• PHOTON: Hierarchical Autoregressive Modeling for Lightspeed and Memory-Efficient Language Generation
• FEM-Bench: A Structured Scientific Reasoning Benchmark for Evaluating Code-Generating LLMs
• Stabilizing Multimodal Autoencoders: A Theoretical and Empirical Analysis of Fusion Strategies
• Bridging Efficiency and Safety: Formal Verification of Neural Networks with Early Exits
• Generalization of RLVR Using Causal Reasoning as a Testbed
• TS-Arena Technical Report -- A Pre-registered Live Forecasting Platform
• Subgroup Discovery with the Cox Model
• Improving Matrix Exponential for Generative AI Flows: A Taylor-Based Approach Beyond Paterson--Stockmeyer
• Symbolic regression for defect interactions in 2D materials
• GraphFire-X: Physics-Informed Graph Attention Networks and Structural Gradient Boosting for Building-Scale Wildfire Preparedness at the Wildland-Urban Interface
• FedMPDD: Communication-Efficient Federated Learning with Privacy Preservation Attributes via Projected Directional Derivative
• Defending against adversarial attacks using mixture of experts
• Memory-Efficient Acceleration of Block Low-Rank Foundation Models on Resource Constrained GPUs
• Robustness Certificates for Neural Networks against Adversarial Attacks
• From GNNs to Symbolic Surrogates via Kolmogorov-Arnold Networks for Delay Prediction
• Time-Efficient Evaluation and Enhancement of Adversarial Robustness in Deep Neural Networks
• DiEC: Diffusion Embedded Clustering
• Towards a General Framework for Predicting and Explaining the Hardness of Graph-based Combinatorial Optimization Problems using Machine Learning and Association Rule Mining
• RevFFN: Memory-Efficient Full-Parameter Fine-Tuning of Mixture-of-Experts LLMs with Reversible Blocks
• Guardrailed Elasticity Pricing: A Churn-Aware Forecasting Playbook for Subscription Strategy
• A Multi-fidelity Double-Delta Wing Dataset and Empirical Scaling Laws for GNN-based Aerodynamic Field Surrogate
• Solving Functional PDEs with Gaussian Processes and Applications to Functional Renormalization Group Equations
• ReACT-Drug: Reaction-Template Guided Reinforcement Learning for de novo Drug Design
• Can Agentic AI Match the Performance of Human Data Scientists?
• Generalization of Diffusion Models Arises with a Balanced Representation Space
• Generalised Linear Models in Deep Bayesian RL with Learnable Basis Functions
• CoSeNet: A Novel Approach for Optimal Segmentation of Correlation Matrices
• LLM Swiss Round: Aggregating Multi-Benchmark Performance via Competitive Swiss-System Dynamics
• Understanding Scaling Laws in Deep Neural Networks via Feature Learning Dynamics
• Shared Representation Learning for High-Dimensional Multi-Task Forecasting under Resource Contention in Cloud-Native Backends
• A Mechanistic Analysis of Transformers for Dynamical Systems
• STLDM: Spatio-Temporal Latent Diffusion Model for Precipitation Nowcasting
• MODE: Multi-Objective Adaptive Coreset Selection
• BALLAST: Bandit-Assisted Learning for Latency-Aware Stable Timeouts in Raft
• A Unified Framework for EEG Seizure Detection Using Universum-Integrated Generalized Eigenvalues Proximal Support Vector Machine
• Analytic and Variational Stability of Deep Learning Systems
• MiST: Understanding the Role of Mid-Stage Scientific Training in Developing Chemical Reasoning Models
• Improving the Convergence Rate of Ray Search Optimization for Query-Efficient Hard-Label Attacks
• Model Merging via Multi-Teacher Knowledge Distillation
• Transcriptome-Conditioned Personalized De Novo Drug Generation for AML Using Metaheuristic Assembly and Target-Driven Filtering
• Learning to Solve PDEs on Neural Shape Representations
• Does the Data Processing Inequality Reflect Practice? On the Utility of Low-Level Tasks
• Measuring all the noises of LLM Evals
• Uncovering Patterns of Brain Activity from EEG Data Consistently Associated with Cybersickness Using Neural Network Interpretability Maps
• Uncovering Competency Gaps in Large Language Models and Their Benchmarks
• Graph Neural Networks for Source Detection: A Review and Benchmark Study
• Fast and Exact Least Absolute Deviations Line Fitting via Piecewise Affine Lower-Bounding
• Diffusion Models in Simulation-Based Inference: A Tutorial Review
• Mechanism-Based Intelligence (MBI): Differentiable Incentives for Rational Coordination and Guaranteed Alignment in Multi-Agent Systems
• Real-World Adversarial Attacks on RF-Based Drone Detectors
• AI-Driven Green Cognitive Radio Networks for Sustainable 6G Communication
• AgentMath: Empowering Mathematical Reasoning for Large Language Models via Tool-Augmented Agent
• TrashDet: Iterative Neural Architecture Search for Efficient Waste Detection
• A Physics Informed Neural Network For Deriving MHD State Vectors From Global Active Regions Observations
• TokSuite: Measuring the Impact of Tokenizer Choice on Language Model Behavior
• NULLBUS: Multimodal Mixed-Supervision for Breast Ultrasound Segmentation via Nullable Global-Local Prompts
• Weighted MCC: A Robust Measure of Multiclass Classifier Performance for Observations with Individual Weights
• Context-Sensitive Abstractions for Reinforcement Learning with Parameterized Actions
• CHAMMI-75: pre-training multi-channel models with heterogeneous microscopy images
• Nemotron 3 Nano: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
• NVIDIA Nemotron 3: Efficient and Open Intelligence
• Better Call Graphs: A New Dataset of Function Call Graphs for Malware Classification
• Architectural Trade-offs in Small Language Models Under Compute Constraints
• Clever Hans in Chemistry: Chemist Style Signals Confound Activity Prediction on Public Benchmarks
• AirGS: Real-Time 4D Gaussian Streaming for Free-Viewpoint Video Experiences
• MultiMind at SemEval-2025 Task 7: Crosslingual Fact-Checked Claim Retrieval via Multi-Source Alignment
• Deadline-Aware Online Scheduling for LLM Fine-Tuning with Spot Market Predictions
• GenTSE: Enhancing Target Speaker Extraction via a Coarse-to-Fine Generative Language Model
• Parameter-Efficient Neural CDEs via Implicit Function Jacobians
• Learning Evolving Latent Strategies for Multi-Agent Language Systems without Model Fine-Tuning