Ax Jiaming Hu, Jiamu Bai, Haoyu Wang, Debarghya Mukherjee, Ioannis Ch. Paschalidis 5/7/2026

Towards General Preference Alignment: Diffusion Models at Nash Equilibrium

Preference alignment for text-to-image diffusion models using game-theoretic approach at Nash equilibrium instead of reward modeling.

Ax Akiyoshi Tomihari, Issei Sato 5/7/2026

Power Distribution Bridges Sampling, Self-Reward RL, and Self-Distillation

Analysis connecting power sampling, self-reward RL, and self-distillation for improving LLM reasoning and performance.

Ax Jan Marco Ruiz de Vargas, Fabian Raisch, Zoltan Nagy, Pierre Pinson, Christoph Goebel 5/7/2026

Counter-Dyna: Data-Efficient RL-Based HVAC Control using Counterfactual Building Models

Model-based reinforcement learning for data-efficient HVAC control in buildings using counterfactual building models.

Ax Xinyi Li, Ming Li, Lu Bai, Lixin Cui, Feilong Cao, Ke Lv, Yunliang Jiang, Pietro Li\`o 5/7/2026

HeterSEED: Semantics-Structure Decoupling for Heterogeneous Graph Learning under Heterophily

HeterSEED decouples semantics from structure in heterogeneous graph learning to handle heterophilic graphs with dissimilar connected nodes.

Ax Chengyi Nie, Nian Si, Zijie Zhou 5/7/2026

A Queueing-Theoretic Framework for Stability Analysis of LLM Inference with KV Cache Memory Constraints

Queueing-theoretic framework for analyzing LLM inference stability under KV cache memory constraints and computation bottlenecks.

Ax Guangsheng Bao, Hongbo Zhang, Han Cui, Yanbin Zhao, Yue Zhang 5/7/2026

FAAST: Forward-Only Associative Learning via Closed-Form Fast Weights for Test-Time Supervised Adaptation

FAAST enables forward-only test-time adaptation by analytically compiling labeled examples into fast weights without backpropagation overhead.

Ax Jinbin Bai, Yu Lei, Qingyu Shi, Aosong Feng, Yi Xin, Zhuoran Zhao, Fei Shen, Kaidong Yu, Jason Li 5/7/2026

Threshold-Guided Optimization for Visual Generative Models

Threshold-guided optimization for visual generative models using implicit reward comparisons to enable scalar rating-based preference alignment.

Ax Ye Su, Longlong Zhao, Diego Garcia-Gil, Jipeng Guo, Gangchun Zhang, Jinxin Chen, Jinsong Chen 5/7/2026

ITBoost: Information-Theoretic Trust for Robust Boosting

ITBoost improves gradient boosting robustness to noisy labels by using information-theoretic evaluation of sample reliability.

Ax Lirui Luo, Guoxi Zhang, Hongming Xu, Cong Fang, Qing Li 5/7/2026

SPHERE: Mitigating the Loss of Spectral Plasticity in Mixture-of-Experts for Deep Reinforcement Learning

SPHERE mitigates plasticity loss in mixture-of-experts networks for continual deep reinforcement learning by preserving spectral properties.

Ax Jaewook Kim, Hyeoncheol Kim 5/7/2026

Ensuring Reliability in Programming Knowledge Tracing: A Re-evaluation of Attention-augmented Models and Experimental Protocols

Re-evaluation of attention-based programming knowledge tracing models reveals sensitivity to implementation details and experimental protocols.

Ax Zhikai Li, Zhen Dong, Xuewen Liu, Jing Zhang, Qingyi Gu 5/7/2026

OSAQ: Outlier Self-Absorption for Accurate Low-bit LLM Quantization

OSAQ addresses weight outliers in post-training LLM quantization to reduce model size and inference latency through selective absorption.

Ax Leon Witt, Togrul Abbasli, Kentaroh Toyoda, Wojciech Samek, Lucy Klinger 5/7/2026

Knowledge-Free Correlated Agreement for Incentivizing Federated Learning

KFCA mechanism rewards federated learning client contributions without ground truth labels, evaluated on LLM adapter tuning and PCB inspection tasks.

Ax Omkar B Shende, Marcello Traiola, Gayathri Ananthanarayanan 5/7/2026

AxMoE: Characterizing the Impact of Approximate Multipliers on Mixture-of-Experts DNN Architectures

AxMoE characterizes the impact of approximate multipliers on mixture-of-experts DNN architectures for efficient edge inference.

Ax Wu-Yuin Hwang, Nur Alif Ilyasa, Muhammad Irfan Luthfi, Yuniar Indrihapsari 5/7/2026

Cognitive Twins: Investigating Personalized Thinking Model Building and Its Performance Enhancement with Human-in-the-Loop

Personalized Thinking Model for AI-supported education capturing learner behavior, cognition, and metacognition in hierarchical structure.

Ax Matan Pagi, Zohar Sorek 5/7/2026

Bilinear Mamba-Koopman Neural MPC for Varying Dynamics

Bilinear Mamba-Koopman Neural MPC for model predictive control with time-varying dynamics from historical data.

Ax Zhenchao Sun, Shuai Ma, Ping Lu, Chongyang Tao 5/7/2026

Unsat Core Prediction through Polarity-Aware Representation Learning over Clause-Literal Hypergraphs

Uses graph neural networks and hypergraph representation learning for unsat-core prediction in Boolean satisfiability problems.

Ax Taiga Hayashi, Yuya Seki, Kotaro Terada, Yosuke Mukasa, Shuta Kikuchi, Shu Tanaka 5/7/2026

Improving FMQA via Initial Training Data Design Considering Marginal Bit Coverage in One-Hot Encoding

Improves FMQA black-box optimization through better initial training data design considering marginal bit coverage in one-hot encoding.

Ax Junxiang Wu, Zhiqiang Kou, Hongwei Zeng, Wenke Huang, Biao Liu, Hanlin Gu, Yuheng Jia, Di Jiang, Yang Liu, Xin Geng, Qiang Yang 5/7/2026

Trustworthy Federated Label Distribution Learning under Annotation Quality Disparity

Federated label distribution learning approach handling heterogeneous annotation quality across clients in privacy-sensitive settings.

Ax Yifan F. Zhang, Fangjun Hu, Guangkuo Liu, Mert Okyay, Xun Gao 5/7/2026

Concurrence of Symmetry Breaking and Nonlocality Phase Transitions in Diffusion Models

Analyzes phase transitions in diffusion models through symmetry breaking and nonlocality during generation dynamics.

Ax Yizheng Wang, Mohammad Sadegh Eshaghi, Xiaoying Zhuang, Timon Rabczuk, Yinghua Liu 5/7/2026

Replay-Based Continual Learning for Physics-Informed Neural Operators

Continual learning approach for physics-informed neural operators using replay to address out-of-distribution performance degradation.

Ax Moshe Eliasof, Krishna Sri Ipsit Mantri, Beatrice Bevilacqua, Bruno Ribeiro, Carola-Bibiane Sch\"onlieb 5/7/2026

Bridging Input Feature Spaces Towards Graph Foundation Models

ALL-IN method enables graph neural network transferability across datasets with different input feature spaces, enabling graph foundation models.

Ax Soyoung park, Hwanjun Song, Sungsu Lim 5/7/2026

Quantile-Free Uncertainty Quantification in Graph Neural Networks

QpiGNN introduces quantile-free uncertainty quantification for graph neural networks without requiring costly resampling or post-hoc calibration.

Ax Huatian Zhang, Zhendong Mao, Lei Zhang, Yongdong Zhang 5/7/2026

Uncertainty-Aware Exploratory Direct Preference Optimization for Multimodal Large Language Models

Uncertainty-aware DPO method for reducing hallucination in multimodal LLMs by allocating fine-grained supervision to vision tokens.

Ax Dominik Dahlem, Diego Maniloff, Mac Misiura 5/7/2026

Self-Attention as Transport: Limits of Symmetric Spectral Diagnostics

Theoretical analysis proving limitations of symmetric spectral methods for diagnosing attention failures in language models.

Ax Albert F. Modenbach 5/7/2026

A geometric relation of the error introduced by sampling a language model's output distribution to its internal state

Geometric analysis of sampling errors in language models, showing curvature of token embedding geometry correlates with semantic properties.

Ax Santosh Premi Adhikari, Radu Timofte, Dmitry Ignatov 5/7/2026

Delta-Based Neural Architecture Search: LLM Fine-Tuning via Code Diffs

Delta-Code Generation: LLMs generate compact unified diffs to refine baseline architectures rather than complete models from scratch.

Ax Xinyan Han, Yan Lu, Xiaoyu Lin, Yuanyuan Jiang, Yuanrui Wang, Xuanyue Li, Wenchao Zou, Xingxuan Zhang 5/7/2026

Breaking the Quality-Privacy Tradeoff in Tabular Data Generation via In-Context Learning

In-context learning approach for tabular data generation balancing quality and privacy without model-specific training.

Ax Xiyan Fu, Wei Liu 5/7/2026

Reinforcement Learning for Compositional Generalization with Outcome-Level Optimization

Reinforcement learning approach for compositional generalization using outcome-level optimization instead of token-level training.

Ax Yu-Xuan Tao, Tian-Zuo Wang, Zhi-Hua Zhou 5/7/2026

Order-based Rehearsal Learning

Learning method for avoiding undesired futures using order structure instead of graph structure in forecasting scenarios.

Ax Han Wang, Jintao Zhang, Kai Jiang, Haoxu Wang, Jianfei Chen, Jun Zhu 5/7/2026

KernelBench-X: A Comprehensive Benchmark for Evaluating LLM-Generated GPU Kernels

KernelBench-X benchmark evaluating LLM-generated GPU kernels across 176 tasks, identifying where LLM kernel generation capabilities break down.

Ax Song Yu, Li Li, Wenwen Zhao, Zhisheng Yang 5/7/2026

EP-GRPO: Entropy-Progress Aligned Group Relative Policy Optimization with Implicit Process Guidance

EP-GRPO algorithm addressing credit assignment failures in reinforcement learning for LLM reasoning, improving upon Group Relative Policy Optimization.

Ax Antonin Berthon, Nicolas Astorga, Mihaela van der Schaar 5/7/2026

Skill Neologisms: Towards Skill-based Continual Learning

Method for extending LLM capabilities to new skills using soft tokens (skill neologisms) to avoid catastrophic forgetting while maintaining expressiveness beyond context limits.

Ax Ilias Triantafyllopoulos, Young-Min Cho, Ren Tao, Miranda Muqing Miao, Sunny Rai, Lyle Ungar, Sharath Chandra Guntuku, Neville Ryant, Jo\~ao Sedoc 5/7/2026

Conceptors for Semantic Steering

Conceptor-based steering method for controlling LLM behavior using soft projection matrices that preserve multidimensional concept subspaces.

Ax Senkang Hu, Yong Dai, Xudong Han, Zhengru Fang, Yuzhi Zhao, Sam Tak Wu Kwong, Yuguang Fang 5/7/2026

Self-Induced Outcome Potential: Turn-Level Credit Assignment for Agents without Verifiers

Self-Induced Outcome Potential method for turn-level credit assignment in long-horizon LLM agents without process-level verifiers.

Ax Anastasis Kratsios, A. Martina Neuman, Philipp Petersen 5/7/2026

Adaptivity Under Realizability Constraints: Comparing In-Context and Agentic Learning

Theoretical comparison of in-context learning vs agentic learning for task approximation under neural network realizability constraints.

Ax Michael Lanier, Luise Ge, Sastry Kompella, Yevgeniy Vorobeychik 5/7/2026

Learned Neighbor Trust for Collaborative Deployment in Model-Agnostic Decentralized Learning

Training method for decentralized learning where heterogeneous nodes learn to compose predictions effectively at inference time.

Ax Shawn Ray 5/7/2026

Graph-SND: Sparse Aggregation for Behavioral Diversity in Multi-Agent Reinforcement Learning

Graph-SND metric for measuring behavioral diversity in multi-agent RL using sparse graph aggregation instead of complete pairwise distances.

Ax Xing Ma, Yangjie Zhou, Wu Sun, Zihan Liu, Jingwen Leng, Yun Lin, Shixuan Sun, Minyi Guo, Jin Song Dong 5/7/2026

CuBridge: An LLM-Based Framework for Understanding and Reconstructing High-Performance Attention Kernels

CuBridge uses LLMs to understand and generate high-performance CUDA attention kernels with improved correctness and efficiency.

Ax Kejun Liu 5/7/2026

The Predictive-Causal Gap: An Impossibility Theorem and Large-Scale Neural Evidence

Empirical study showing optimal predictive encoders fail at causal fidelity, with analysis of 2695 neural networks on linear-Gaussian dynamics.

Ax Xin Yu, Liuchen Liao, Yiwen Zhang, Yingchen Yu, Lingzhou Xue, Qinzhen Guo 5/7/2026

Preference-Based Self-Distillation: Beyond KL Matching via Reward Regularization

Self-distillation method for on-policy LLM training using preference-based reward regularization beyond KL matching.

Ax Xiaofan Xia, Qin Li, Wenlong Mou 5/7/2026

Provable imitation learning for control of instability in partially-observed Vlasov--Poisson equations

Imitation learning approach for plasma stabilization control in Vlasov-Poisson systems with partial state observability constraints.

Ax Andrea Napoli, Paul White 5/7/2026

Order Matters: Improving Domain Adaptation by Reordering Data

ORDERED algorithm for unsupervised domain adaptation that reduces variance in discrepancy estimation through optimal data reordering.

Ax Andreas Pattichis, Constantine Dovrolis 5/7/2026

Continual Knowledge Updating in LLM Systems: Learning Through Multi-Timescale Memory Dynamics

Memini system for continual knowledge updating in LLMs using multi-timescale memory dynamics inspired by biological memory mechanisms.

Ax Harin Lee, Min-hwan Oh 5/7/2026

Unified Framework of Distributional Regret in Multi-Armed Bandits and Reinforcement Learning

Theoretical framework for analyzing regret distribution in multi-armed bandits and episodic RL with probabilistic guarantees across confidence levels.

Ax Tianshu Zhu, Wenyu Zhang, Xiaoying Zuo, Lun Tian, Haotian Zhao, Yucheng Zeng, Jingnan Gu, Daxiang Dong, Jianmin Wu, Dawei Yin, Dou Shen 5/7/2026

Rollout Pass-Rate Control: Steering Binary-Reward RL Toward Its Most Informative Regime

RL method for improving sample efficiency in agentic tasks by controlling pass-rate to optimal 50% for maximizing reward signal informativeness.

Ax Mariia Seleznova 5/7/2026

How Long Does Infinite Width Last? Signal Propagation in Long-Range Linear Recurrences

Theoretical study of signal propagation in finite-width linear recurrent models, analyzing approximation accuracy as sequence depth and width grow jointly.

Ax Marco Rando, Samuel Vaiter 5/7/2026

On the Hardness of Junking LLMs

Analysis of jailbreak attack difficulty on LLMs, showing random search effectiveness challenges assumptions about prompt structure necessity.

Ax Dan Wilson, Mohamed Akrout 5/7/2026

Low-Cost Black-Box Detection of LLM Hallucinations via Dynamical System Prediction

Low-cost black-box method to detect LLM hallucinations by treating model as dynamical system and analyzing embedding manifolds.

Ax Freyaa Chawla, Ahan Chawla, Rishi Singh, Joe Germino, Grigorii Khvatskii 5/7/2026

Human-AI Co-Mentorship in Project-Based Learning: A Case Study in Financial Forecasting

Case study of AI tools assisting high school students in financial forecasting research project with human-AI co-mentorship.

Ax Alper Y{\i}ld{\i}r{\i}m 5/7/2026

Superposition Is Not Necessary: A Mechanistic Interpretability Analysis of Transformer Representations for Time Series Forecasting

Mechanistic interpretability study of transformer time series forecasting using sparse autoencoders, questioning superposition hypothesis.