3/31/2026
Efficient Domain Adaptation for Text Line Recognition via Decoupled Language Models
Modular transformer approach for efficient domain adaptation in optical character recognition with reduced computational requirements.
3/31/2026
Beyond the Answer: Decoding the Behavior of LLMs as Scientific Reasoners
Research on how LLMs perform scientific reasoning tasks and how prompting affects their internal reasoning processes.
3/31/2026
A Deep Reinforcement Learning Framework for Closed-loop Guidance of Fish Schools via Virtual Agents
Deep reinforcement learning framework using PPO to train virtual agents for guiding fish school collective motion.
3/31/2026
A Multi-Agent Rhizomatic Pipeline for Non-Linear Literature Analysis
Multi-agent pipeline for literature analysis using Deleuzian ontology to identify non-linear patterns in research landscapes.
3/31/2026
Kernel-Smith: A Unified Recipe for Evolutionary Kernel Optimization
Framework for evolutionary GPU kernel optimization using evaluation-driven agent and evolutionary techniques for operator generation.
3/31/2026
The Scaffold Effect: How Prompt Framing Drives Apparent Multimodal Gains in Clinical VLM Evaluation
Analysis of prompt framing artifacts in vision-language model evaluation on clinical neuroimaging tasks.
3/31/2026
From Simulation to Deep Learning: Survey on Network Performance Modeling Approaches
Survey of network performance modeling approaches comparing traditional simulation with deep learning methods.
3/31/2026
$R_{dm}$: Re-conceptualizing Distribution Matching as a Reward for Diffusion Distillation
Diffusion distillation approach using reinforcement learning to improve student model performance beyond teacher anchoring.
3/31/2026
Towards a Medical AI Scientist
Medical AI Scientist system that autonomously generates hypotheses, conducts experiments, and writes manuscripts in clinical medicine.
3/31/2026
Constructing Composite Features for Interpretable Music-Tagging
Genetic programming pipeline for automatically evolving interpretable composite features for music tagging tasks.
3/31/2026
On-the-fly Repulsion in the Contextual Space for Rich Diversity in Diffusion Transformers
Method for improving diversity in text-to-image diffusion transformers through contextual space repulsion to address typicality bias.
3/31/2026
Less is More: Rethinking Few-Shot Learning and Recurrent Neural Nets
Study on few-shot learning and RNNs applying asymptotic equipartition property from information theory to machine learning.
3/31/2026
Convergence of the Inexact Langevin Algorithm in KL Divergence with Application to Score-based Generative Models
Theoretical analysis of inexact Langevin algorithm convergence for score-based generative models with KL divergence guarantees.
3/31/2026
Continual Graph Learning: A Survey
Survey on continual graph learning covering incremental learning from streaming graph data with experience and generative replay approaches.
3/31/2026
Correcting Auto-Differentiation in Neural-ODE Training
Mathematical analysis of auto-differentiation reliability in neural-ODE training with high-order numerical methods.
3/31/2026
Learning Expressive Priors for Generalization and Uncertainty Estimation in Neural Networks
Novel prior learning method for neural networks using structured posteriors to improve generalization and uncertainty estimation.
3/31/2026
Unichain and Aperiodicity are Sufficient for Asymptotic Optimality of Average-Reward Restless Bandits
Proves asymptotic optimality of new restless bandit policies with O(1/√N) gap under unichain and aperiodicity conditions.
3/31/2026
Learning the Model While Learning Q: Finite-Time Sample Complexity of Online SyncMBQ
Theoretical analysis of sample complexity for model-based Q-learning, establishing finite-time convergence bounds for model-learning algorithms.
3/31/2026
Remedying uncertainty representations in visual inference through Explaining-Away Variational Autoencoders
Paper proposing Explaining-Away Variational Autoencoders to improve uncertainty representations in deep generative models for visual inference tasks.
3/31/2026
Recent Advances of Multimodal Continual Learning: A Comprehensive Survey
Survey of multimodal continual learning methods that enable models to learn from new data across multiple modalities while retaining previous knowledge without catastrophic forgetting.
3/31/2026
Transformers learn variable-order Markov chains in-context
Transformers learn variable-order Markov chains in-context with finite-sample accuracy analysis using context-tree weighting.
3/31/2026
Gradient Compression Beyond Low-Rank: Wavelet Subspaces Compact Optimizer States
Wavelet subspace compression for optimizer states reduces memory during LLM training, improving upon low-rank approaches.
3/31/2026
Shifting Perspectives: Steering Vectors for Robust Bias Mitigation in LLMs
Steering vectors applied to LLM activations for bias mitigation across social dimensions like age, gender, and race.
3/31/2026
Large Language Models for Computer-Aided Design: A Survey
Survey of LLM integration with Computer-Aided Design tools, covering applications in 3D modeling and design workflows.
3/31/2026
Context parroting: A simple but tough-to-beat baseline for foundation models in scientific machine learning
Foundation models for time-series prediction often use simple parroting strategies rather than learning physics, revealing shared failure modes.
3/31/2026
Structured Agent Distillation for Large Language Model
Structured Agent Distillation compresses large LLM-based ReAct agents into smaller models while preserving reasoning and action consistency.
3/31/2026
CoDec: Prefix-Shared Decoding Kernel for LLMs
CoDec kernel optimizes LLM decoding by sharing prefix computation across multiple prompts to reduce memory-intensive KV cache access.
3/31/2026
MicroMix: Efficient Mixed-Precision Quantization with Microscaling Formats for Large Language Models
MicroMix: mixed-precision quantization method using microscaling formats for efficient LLM inference on NVIDIA Blackwell hardware.
3/31/2026
Forgetting: A New Mechanism Towards Better Large Language Model Fine-tuning
Novel fine-tuning mechanism for LLMs that addresses data quality/volume issues through controlled forgetting to improve domain adaptation.
3/31/2026
PENGUIN: Enhancing Transformer with Periodic-Nested Group Attention for Long-term Time Series Forecasting
PENGUIN: Transformer variant with periodic-nested group attention mechanism for improved long-term time series forecasting.
3/31/2026
Initialization Schemes for Kolmogorov-Arnold Networks: An Empirical Study
Empirical study of initialization schemes for Kolmogorov-Arnold Networks, proposing theory-driven approaches to improve training of spline-based KANs.
3/31/2026
No Need for Learning to Defer? A Training Free Deferral Framework to Multiple Experts through Conformal Prediction
Training-free framework for deferring predictions to multiple experts using conformal prediction without retraining.
3/31/2026
ReTrack: Data Unlearning in Diffusion Models through Redirecting the Denoising Trajectory
ReTrack enables data unlearning in diffusion models via importance sampling to remove memorized training data influence.
3/31/2026
Asynchronous Policy Gradient Aggregation for Efficient Distributed Reinforcement Learning
Algorithms for distributed RL with policy gradients under asynchronous parallel computation and communication.
3/31/2026
LLM as an Algorithmist: Enhancing Anomaly Detectors via Programmatic Synthesis
Uses LLMs to programmatically synthesize anomaly detectors for tabular data without direct processing of raw data for privacy.
3/31/2026
Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models
ACE framework evolves context for self-improving LLM agents, addressing brevity bias and context collapse in iterative refinement.
3/31/2026
Mitigating Premature Exploitation in Particle-based Monte Carlo for Inference-Time Scaling
Mitigates premature exploitation in particle filtering for inference-time scaling of language models using process reward models.
3/31/2026
TabPFN-Wide: Continued Pre-Training for Extreme Feature Counts
TabPFN-Wide extends prior-data fitted networks for tabular data with extreme feature counts in biomedicine applications.
3/31/2026
Constraints-of-Thought: A Framework for Constrained Reasoning in Language-Model-Guided Search
Constraints-of-Thought framework enables LLMs to perform constrained multi-step reasoning while satisfying symbolic constraints and user intent.
3/31/2026
PANTHER: Generative Pretraining Beyond Language for Sequential User Behavior Modeling
PANTHER applies generative pretraining to model user behavior sequences beyond language, using multi-dimensional action attributes.
3/31/2026
Learning When Not to Learn: Risk-Sensitive Abstention in Bandits with Unbounded Rewards
Bandit algorithm for high-stakes sequential decision-making that learns when to abstain from actions with irreparable consequences.
3/31/2026
Dense and Diverse Goal Coverage in Multi Goal Reinforcement Learning
RL algorithm for learning policies that maximize return while inducing dispersed state distributions across multiple reward sources.
3/31/2026
Decomposable Neuro Symbolic Regression
Transformer-based symbolic regression method for discovering interpretable mathematical expressions from observed data.
3/31/2026
From Exploration to Exploitation: A Two-Stage Entropy RLVR Approach for Noise-Tolerant MLLM Training
Two-stage entropy approach for noise-tolerant multimodal LLM training using reinforcement learning with verifiable rewards.
3/31/2026
Object-Centric World Models for Causality-Aware Reinforcement Learning
Object-centric world models for reinforcement learning using decomposed representations to improve sample efficiency in multi-object environments.
3/31/2026
UniGame: Turning a Unified Multimodal Model Into Its Own Adversary
UniGame addresses inconsistency in unified multimodal models between understanding and generation through adversarial framework.
3/31/2026
Single-Round Scalable Analytic Federated Learning
SAFLe framework enabling scalable non-linear federated learning in a single round with heterogeneous data distribution invariance.
3/31/2026
Prototype-Based Semantic Consistency Alignment for Domain Adaptive Retrieval
Domain adaptive retrieval using prototype-based semantic consistency alignment to transfer knowledge from labeled to unlabeled domains.
3/31/2026
Measuring all the noises of LLM Evals
Research on measuring noise in LLM evaluations using statistical methods to separate signal from noise in prediction, data, and combined noise.
3/31/2026
Electricity Price Forecasting: Bridging Linear Models, Neural Networks and Online Learning
Day-ahead electricity price forecasting combining linear models, neural networks and online learning for volatile market prediction.