4/29/2026
Subspace Optimization for Efficient Federated Learning under Heterogeneous Data
Subspace optimization method for federated learning addressing non-IID data heterogeneity with reduced communication and memory overhead.
4/29/2026
EvoTSC: Evolving Feature Learning Models for Time Series Classification via Genetic Programming
Genetic programming approach for automatically evolving lightweight feature learning models for time series classification with limited labeled data.
4/29/2026
Dyna-Style Safety Augmented Reinforcement Learning: Staying Safe in the Face of Uncertainty
Safe RL algorithm combining learned dynamics models and uncertainty-aware safety filters for safe exploration in high-dimensional systems.
4/29/2026
On Halting vs Converging in Recurrent Graph Neural Networks
Comparative study of recurrent graph neural network variants: converging, output-converging, and halting RGNNs for iterative message-passing.
4/29/2026
Towards interpretable AI with quantum annealing feature selection
Quantum annealing-based feature selection method for interpreting convolutional neural networks in critical applications.
4/29/2026
Diverse Image Priors for Black-box Data-free Knowledge Distillation
Black-box data-free knowledge distillation using diverse image priors to transfer knowledge from teacher to student models without training data access.
4/29/2026
Barriers to Universal Reasoning With Transformers (And How to Overcome Them)
Examines whether transformers can generalize chain-of-thought reasoning beyond training trace lengths, using theoretical frameworks for length generalization.
4/29/2026
Investigation into In-Context Learning Capabilities of Transformers
Empirical investigation of transformer in-context learning capabilities, characterizing scaling behavior and conditions for ICL success on unseen tasks.
4/29/2026
When Errors Can Be Beneficial: A Categorization of Imperfect Rewards for Policy Gradient
Analyzes how imperfect proxy rewards used in RL training of LLMs vary in harmfulness, proposing a categorization framework for understanding which reward errors are beneficial vs harmful.
4/29/2026
Conditional misalignment: common interventions can hide emergent misalignment behind contextual triggers
Studies how interventions reducing emergent misalignment in finetuned LLMs may hide problematic behaviors behind contextual triggers rather than fully mitigating them.
4/29/2026
TSN-Affinity: Similarity-Driven Parameter Reuse for Continual Offline Reinforcement Learning
TSN-Affinity enables parameter reuse for continual offline reinforcement learning across sequential tasks without catastrophic forgetting.
4/29/2026
How Fast Should a Model Commit to Supervision? Training Reasoning Models on the Tsallis Loss Continuum
Training reasoning models using Tsallis loss continuum to balance exploitation and density-estimation in low-success reinforcement learning settings.
4/29/2026
A systematic literature Review for Transformer-based Software Vulnerability detection
Systematic literature review of transformer-based models for automated software vulnerability detection.
4/29/2026
FGDM: Reasoning Aware Multi-Agentic Framework for Software Bug Detection using Chain of Thought and Tree of Thought Prompting
FGDM uses multi-agentic framework with LLMs, CoT and ToT prompting for automated software bug detection in complex codebases.
4/29/2026
Learning Illumination Control in Diffusion Models
Open-source reproducible pipeline for learning illumination control in diffusion models with data engine and released code.
4/29/2026
VibeToken: Scaling 1D Image Tokenizers and Autoregressive Models for Dynamic Resolution Generations
VibeToken: resolution-agnostic 1D Transformer-based image tokenizer enabling efficient autoregressive image synthesis at arbitrary resolutions.
4/29/2026
VISION-SLS: Safe Perception-Based Control from Learned Visual Representations via System Level Synthesis
Control method from RGB images using learned visual representations with safety guarantees via system-level synthesis.
4/29/2026
An analysis of sensor selection for fruit picking with suction-based grippers
Analysis of sensor selection strategies for robotic fruit harvesting with suction grippers, focusing on pick state estimation.
4/29/2026
Data-Driven Hamiltonian Reduction for Superconducting Qubits via Meta-Learning
Meta-learning framework for fast online adaptation of Hamiltonian models in superconducting quantum processors using hardware measurements.
4/29/2026
Large Language Models Explore by Latent Distilling
Exploratory Sampling decoding method for LLMs that encourages semantic diversity in generated responses during test-time scaling.
4/29/2026
CAN-QA: A Question-Answering Benchmark for Reasoning over In-Vehicle CAN Traffic
Question-answering benchmark for reasoning over in-vehicle CAN traffic data, formulating intrusion detection as a reasoning task.
4/29/2026
A Finite Time Analysis of Thompson Sampling for Bayesian Optimization with Preferential Feedback
Thompson Sampling approach for Bayesian optimization using pairwise preference feedback instead of scalar scores, with theoretical analysis.
4/29/2026
Spark Policy Toolkit: Semantic Contracts and Scalable Execution for Policy Learning in Spark
Spark Policy Toolkit providing Spark-native primitives for scalable policy learning with semantic contracts and vectorized inference.
4/29/2026
Frontier Coding Agents Can Now Implement an AlphaZero Self-Play Machine Learning Pipeline For Connect Four That Performs Comparably to an External Solver
Benchmark measuring AI agents' capability to autonomously implement end-to-end ML pipelines from minimal task descriptions, demonstrated with Connect Four AlphaZero implementation.
4/29/2026
Doing More With Less: Revisiting the Effectiveness of LLM Pruning for Test-Time Scaling
Investigation of LLM pruning effectiveness for test-time scaling, examining how structured pruning impacts reasoning capabilities and model compression.
4/29/2026
Frictive Policy Optimization for LLMs: Epistemic Intervention, Risk-Sensitive Control, and Reflective Alignment
Frictive Policy Optimization framework for LLMs that treats clarification, verification, challenge, redirection, and refusal as explicit control actions to manage epistemic and normative risk.
4/29/2026
How Can Reinforcement Learning Achieve Expert-level Placement?
Reinforcement learning method for chip placement in physical design that learns from expert layouts.
4/29/2026
Making AI-Assisted Grant Evaluation Auditable without Exposing the Model
TEE-based architecture for auditable LLM-assisted decision-making in grant evaluation while protecting model confidentiality.
4/29/2026
BARRED: Synthetic Training of Custom Policy Guardrails via Asymmetric Debate
Framework for training custom policy guardrails for LLMs using synthetic data generation via debate-based refinement.
4/29/2026
Walking Through Uncertainty: An Empirical Study of Uncertainty Estimation for Audio-Aware Large Language Models
Research on uncertainty estimation in audio-aware large language models to reduce hallucinations and overconfident outputs.
4/29/2026
PLMGH: What Matters in PLM-GNN Hybrids for Code Classification and Vulnerability Detection
Empirical study comparing pretrained language models and graph neural networks for code classification and vulnerability detection.
4/29/2026
Residual-loss Anomaly Analysis of Physics-Informed Neural Networks: An Inverse Method for Change-point Detection in Nonlinear Dynamical Systems with Regime Switching
Physics-informed neural networks framework for change-point detection and parameter estimation in nonlinear dynamical systems.
4/29/2026
Adaptive Meta-Learning Stochastic Gradient Hamiltonian Monte Carlo Simulation for Bayesian Updating of Structural Dynamic Models
Meta-learning approach combining MCMC and neural networks for Bayesian structural health monitoring and dynamic model updating.
4/29/2026
Improving Diversity in Black-box Few-shot Knowledge Distillation
Research on black-box few-shot knowledge distillation for compressing neural networks without full teacher access.
4/29/2026
From Soliloquy to Agora: Memory-Enhanced LLM Agents with Decentralized Debate for Optimization Modeling
Agora-Opt: multi-agent LLM framework with decentralized debate and memory for solving optimization modeling problems.
4/29/2026
G-Loss: Graph-Guided Fine-Tuning of Language Models
G-Loss: graph-guided loss function for fine-tuning language models using semi-supervised label propagation on embedding manifolds.
4/29/2026
Explainable AI for Jet Tagging: A Comparative Study of GNNExplainer, GNNShap, and GradCAM for Jet Tagging in the Lund Jet Plane
Comparative study of explainability methods (GNNExplainer, GNNShap, GradCAM) for graph neural networks on particle physics jet tagging.
4/29/2026
Variational Neural Belief Parameterizations for Robust Dexterous Grasping under Multimodal Uncertainty
Research on risk-sensitive grasp planning using variational neural belief parameterizations for dexterous robotic manipulation under uncertainty.
4/29/2026
Toward a Functional Geometric Algebra for Natural Language Semantics
Research proposing geometric algebra framework to address limitations in compositional semantics and interpretability of neural language models.
4/29/2026
Carbon-Taxed Transformers: A Green Compression Pipeline for Overgrown Language Models
Compression pipeline for reducing computational cost and carbon footprint of large language models for software engineering applications.
4/29/2026
Recursive Multi-Agent Systems
RecursiveMAS: recursive multi-agent framework that scales agent collaboration through iterative latent-space refinement for complex reasoning.
4/29/2026
NUBO: A Transparent Python Package for Bayesian Optimization
NUBO: open-source Python package for Bayesian optimization using Gaussian processes and acquisition functions for expensive black-box optimization.
4/29/2026
Bayesian Inverse Transition Learning: Learning Dynamics From Near-Optimal Trajectories
Constraint-based Bayesian method for learning transition dynamics from near-optimal expert trajectories in offline reinforcement learning.
4/29/2026
Soft-TransFormers for Continual Learning
Parameter-efficient continual learning framework using soft masks on frozen pre-trained Transformers for task-specific adaptation.
4/29/2026
Revisiting the Past: Data Unlearning with Model State History
Data unlearning method for LLMs using model state history to efficiently remove influence of problematic training data without full retraining.
4/29/2026
Curriculum-guided multimodal representation learning enables generalizable prediction of nanomaterial-protein interactions
Curriculum-guided multimodal model (CuMMI) for predicting nanomaterial-protein interactions with improved generalization to unseen materials.
4/29/2026
Comparing Data Assimilation and Likelihood-Based Inference on Latent State Estimation in Agent-Based Models
Compares data assimilation and likelihood-based inference methods for estimating latent states in agent-based models against real-world data.
4/29/2026
CHUCKLE -- When Humans Teach AI To Learn Emotions The Easy Way
Curriculum learning approach for emotion recognition using crowdsourced human difficulty assessments instead of heuristic sample difficulty definitions.
4/29/2026
Rethinking Entropy Interventions in RLVR: An Entropy Change Perspective
Analysis of entropy collapse in reinforcement learning with verifiable rewards for LLMs, proposing entropy change perspective for improved training.
4/29/2026
Reinforcement Learning Using known Invariances
Framework for incorporating group symmetries into kernel-based reinforcement learning via symmetry-aware optimistic least-squares value iteration.