2/25/2026
Probing Dec-POMDP Reasoning in Cooperative MARL
Analyzes reasoning capabilities in cooperative multi-agent reinforcement learning under Dec-POMDP framework with partial observability and decentralized coordination.
2/25/2026
Regret-Guided Search Control for Efficient Learning in AlphaZero
Proposes regret-guided search control for AlphaZero to improve learning efficiency through targeted state revisitation in reinforcement learning.
2/25/2026
Transcoder Adapters for Reasoning-Model Diffing
Introduces transcoder adapters to interpret MLP computation differences in reasoning models, comparing Qwen2.5-Math-7B and DeepSeek-R1-Distilled variants.
2/25/2026
From Isolation to Integration: Building an Adaptive Expert Forest for Pre-Trained Model-based Class-Incremental Learning
Semantic-guided expert forest method for class-incremental learning that organizes adapter knowledge through task relationships.
2/25/2026
Extending $\mu$P: Spectral Conditions for Feature Learning Across Optimizers
Extension of maximal update parameterization for hyperparameter transfer across different optimizers in large language model training.
2/25/2026
Does Order Matter : Connecting The Law of Robustness to Robust Generalization
Theoretical work connecting law of robustness to robust generalization in neural networks through Lipschitz constraints.
2/25/2026
Matching Multiple Experts: On the Exploitability of Multi-Agent Imitation Learning
Theoretical analysis of multi-agent imitation learning showing impossibility and hardness results for learning low-exploitable policies offline.
2/25/2026
Localized Dynamics-Aware Domain Adaption for Off-Dynamics Offline Reinforcement Learning
Domain adaptation method for offline reinforcement learning addressing dynamics mismatch through localized similarity matching.
2/25/2026
ProxyFL: A Proxy-Guided Framework for Federated Semi-Supervised Learning
Federated semi-supervised learning framework using proxy guidance to handle data heterogeneity across and within distributed clients.
2/25/2026
Scaling Vision Transformers: Evaluating DeepSpeed for Image-Centric Workloads
Evaluation of DeepSpeed distributed training framework for scaling Vision Transformer models to address computational and memory demands.
2/25/2026
Probing Graph Neural Network Activation Patterns Through Graph Topology
Analysis of Graph Neural Network activation patterns through graph topology and curvature to understand oversmoothing and oversquashing artifacts.
2/25/2026
SOM-VQ: Topology-Aware Tokenization for Interactive Generative Models
Tokenization method combining vector quantization with Self-Organizing Maps to create structured discrete codebooks for interactive generative models.
2/25/2026
SELAUR: Self Evolving LLM Agent via Uncertainty-aware Rewards
Self-evolving LLM agent framework using uncertainty-aware rewards to guide multi-step decision-making and improve learning signal for agent training.
2/25/2026
The Diffusion Duality, Chapter II: $\Psi$-Samplers and Efficient Curriculum
Study on Predictor-Corrector samplers for discrete diffusion models to improve multi-step generation quality beyond ancestral sampling methods.
2/25/2026
Why Pass@k Optimization Can Degrade Pass@1: Prompt Interference in LLM Post-training
Research on Pass@k metric optimization for LLMs showing trade-offs between multi-sample and single-sample inference performance in code generation and reasoning tasks.
2/25/2026
Statistical Query Lower Bounds for Smoothed Agnostic Learning
Statistical query lower bounds for learning halfspaces under Gaussian perturbations in the smoothed agnostic learning model.
2/25/2026
Untied Ulysses: Memory-Efficient Context Parallelism via Headwise Chunking
Untied Ulysses: memory-efficient context parallelism technique for long sequences via headwise chunking in Transformers.
2/25/2026
Learning from Trials and Errors: Reflective Test-Time Planning for Embodied LLMs
Reflective Test-Time Planning for embodied LLMs combining in-action and post-action reflection with test-time scaling for robot task learning.
2/25/2026
Test-Time Training with KV Binding Is Secretly Linear Attention
Analysis revealing that test-time training with KV binding can be expressed as learned linear attention mechanism.
2/25/2026
Benchmarking Distilled Language Models: Performance and Efficiency in Resource-Constrained Settings
Benchmark comparing knowledge-distilled small language models against vanilla and proprietary models for resource-constrained environments.
2/25/2026
Benchmarking Early Deterioration Prediction Across Hospital-Rich and MCI-Like Emergency Triage Under Constrained Sensing
Leakage-aware benchmarking framework for early patient deterioration prediction under realistic emergency triage sensing constraints.
2/25/2026
Regressor-guided Diffusion Model for De Novo Peptide Sequencing with Explicit Mass Control
Deep learning model for de novo peptide sequencing with explicit mass consistency constraints using regressor-guided diffusion.
2/25/2026
Gap-Dependent Bounds for Nearly Minimax Optimal Reinforcement Learning with Linear Function Approximation
Theoretical analysis of gap-dependent regret bounds for reinforcement learning algorithms with linear function approximation.
2/25/2026
An artificial intelligence framework for end-to-end rare disease phenotyping from clinical notes using large language models
LLM-based framework for rare disease phenotyping from clinical notes, extracting and standardizing features to Human Phenotype Ontology terms.
2/25/2026
Circuit Tracing in Vision-Language Models: Understanding the Internal Mechanisms of Multimodal Thinking
Circuit tracing framework for vision-language models using transcoders and attribution graphs to analyze multimodal reasoning mechanisms.
2/25/2026
No One Size Fits All: QueryBandits for Hallucination Mitigation
QueryBandits: model-agnostic contextual bandit framework to mitigate hallucinations in closed-source LLMs through post-hoc detection and mitigation.
2/25/2026
Learning During Detection: Continual Learning for Neural OFDM Receivers via DMRS
Research: Zero-overhead online continual learning framework for deep neural network OFDM receivers adapting to time-varying communication channels.
2/25/2026
Detecting and Mitigating Group Bias in Heterogeneous Treatment Effects
Research on detecting and mitigating group bias in heterogeneous treatment effect predictions when aggregating ML model outputs to subgroups.
2/25/2026
Diffusion Modulation via Environment Mechanism Modeling for Planning
Diffusion-based planning method for offline reinforcement learning with environment mechanism modeling to ensure trajectory consistency.
2/25/2026
Heterogeneity-Aware Client Selection Methodology For Efficient Federated Learning
Heterogeneity-aware client selection for federated learning to improve model accuracy across statistical heterogeneous clients.
2/25/2026
Prior-Agnostic Incentive-Compatible Exploration
Prior-agnostic incentive-compatible exploration in bandit settings with multiple sequential agents.
2/25/2026
ActionEngine: From Reactive to Programmatic GUI Agents via State Machine Memory
ActionEngine: training-free framework for GUI agents using state machine memory to reduce costs, latency, and improve accuracy vs reactive vision-language model approaches.
2/25/2026
Inner Speech as Behavior Guides: Steerable Imitation of Diverse Behaviors for Human-AI coordination
Inner speech guides steerable imitation learning for human-AI coordination, capturing behavioral diversity and non-Markovian human behaviors.
2/25/2026
Stop-Think-AutoRegress: Language Modeling with Latent Diffusion Planning
STAR-LDM integrates latent diffusion planning with autoregressive generation, enabling semantic planning before token commitment.
2/25/2026
Standard Transformers Achieve the Minimax Rate in Nonparametric Regression with $C^{s,\lambda}$ Targets
Theoretical proof that standard Transformers achieve minimax approximation rate for Hölder functions in nonparametric regression.
2/25/2026
Personal Information Parroting in Language Models
Personal information memorization in language models: detector suite for email, phone, IP addresses outperforms existing regex baselines.
2/25/2026
Characterizing Online and Private Learnability under Distributional Constraints via Generalized Smoothness
Learning theory characterizing online and private learnability under distributional constraints via generalized smoothness.
2/25/2026
Knowing the Unknown: Interpretable Open-World Object Detection via Concept Decomposition Model
Interpretable open-world object detection framework using concept decomposition to distinguish known from unknown objects.
2/25/2026
On the Convergence of Stochastic Gradient Descent with Perturbed Forward-Backward Passes
Theoretical analysis of SGD convergence with perturbations in forward-backward passes through sequential operators.
2/25/2026
DANCE: Doubly Adaptive Neighborhood Conformal Estimation
DANCE method for conformal prediction uncertainty quantification using adaptive neighborhood estimation with pre-trained deep learning models.
2/25/2026
Vision-Language Models for Ergonomic Assessment of Manual Lifting Tasks: Estimating Horizontal and Vertical Hand Distances from RGB Video
Vision-language models applied to ergonomic assessment by estimating hand distances from RGB video for NIOSH lifting equation.
2/25/2026
Communication-Inspired Tokenization for Structured Image Representations
Communication-inspired discrete image tokenizer for vision transformers optimized for semantic structure over texture reconstruction.
2/25/2026
SibylSense: Adaptive Rubric Learning via Memory Tuning and Adversarial Probing
SibylSense enables adaptive reward rubric learning for open-ended generation via memory tuning and adversarial probing to prevent reward hacking.
2/25/2026
Don't Ignore the Tail: Decoupling top-K Probabilities for Efficient Language Model Distillation
Tail-aware divergence for language model distillation that decouples top-K probabilities to improve knowledge transfer from teacher to student models.
2/25/2026
DRESS: A Continuous Framework for Structural Graph Refinement
DRESS framework addresses computational complexity of higher-order Weisfeiler-Lehman graph analysis using continuous dynamical systems.
2/25/2026
Functional Continuous Decomposition
Functional Continuous Decomposition framework for non-stationary time-series analysis with parametric optimization and guaranteed continuity.
2/25/2026
SpatiaLQA: A Benchmark for Evaluating Spatial Logical Reasoning in Vision-Language Models
SpatiaLQA benchmark evaluates spatial logical reasoning capabilities in vision-language models for real-world scenarios.
2/25/2026
Some Simple Economics of AGI
Economic analysis of AGI impact on labor, marginal costs, and human verification as binding constraint on growth.
2/25/2026
MIP Candy: A Modular PyTorch Framework for Medical Image Processing
PyTorch framework for medical image processing with support for volumetric data and domain-specific training procedures.
2/25/2026
Is Multi-Distribution Learning as Easy as PAC Learning: Sharp Rates with Bounded Label Noise
Studies multi-distribution learning complexity with bounded label noise to determine if single-task 1/ε rates extend to multi-source settings.