Ax Peter Fernandes, Ria Kanjilal 5/21/2026

GraphRAG on Consumer Hardware: Benchmarking Local LLMs for Healthcare EHR Schema Retrieval

Benchmarks GraphRAG with local LLMs on consumer hardware for healthcare EHR retrieval without cloud dependencies.

Ax Zhiqin Yang, Yonggang Zhang, Wei Xue, Dong Fang, Bo Han, Yike Guo 5/21/2026

Conditional Equivalence of DPO and RLHF: Implicit Assumption, Failure Modes, and Provable Alignment

Proves DPO and RLHF equivalence is conditional on implicit assumptions; identifies failure modes and alignment challenges.

Ax Hanxiang Ren, Pei Zhou, Xunzhe Zhou, Yanchao Yang 5/21/2026

DISC: Decoupling Instruction from State-Conditioned Control via Policy Generation

DISC uses hypernetworks to generate vision-language manipulation policies that decouple instruction processing from state observation.

Ax Yiqi Zhang, Fangzheng Jiao, Tian Tang, Boyu Tian, Hangyu Wang, Qiaoling Chen, Guoteng Wang, Zhen Jiang, Peng Sun, Ping Zhang, Xiaohe Hu, Ziming Liu, Menghao Zhang, Yanmin Jia, Yang You, Siyuan Feng 5/21/2026

PlexRL: Cluster-Level Orchestration of Serviceized LLM Execution for RLVR

PlexRL optimizes cluster-level resource orchestration for reinforcement learning training with verifiable rewards on LLMs.

Ax Ziliang Zhao, Zenan Xu, Shuting Wang, Hongjin Qian, Yan Lei, Minda Hu, Zhao Wang, Shihan Dou, Zhicheng Dou, Pluto Zhou 5/21/2026

PlanningBench: Generating Scalable and Verifiable Planning Data for Evaluating and Training Large Language Models

PlanningBench introduces scalable benchmark generation for evaluating LLM planning capabilities with controllable difficulty levels.

Ax Scott Freitas, Amir Gharib 5/21/2026

GenAI-Driven Threat Detection with Microsoft Security Copilot

Dynamic Threat Detection Agent uses AI to automatically investigate security incidents and adapt detection logic without human intervention.

Ax Hoang Hai Nguyen, Kurt Driessens, Dennis J. N. J. Soemers 5/21/2026

For How Long Should We Be Punching? Learning Action Duration in Fighting Games

RL agent training for fighting games learning variable action durations beyond fixed-frame decisions.

Ax Divyaksh Shukla, Ashutosh Modi 5/21/2026

Calibration vs Decision Making: Revisiting the Reliability Paradox in Unlearned Language Models

Research on calibration vs decision reliability in machine unlearning for language models.

Ax Bole Ma, Jan Eitzinger, Harald Koestler, Gerhard Wellein 5/21/2026

Diagnosing Overhead in Dispatch Operations: Cross-architecture Observatory

Analysis of AlltoAll dispatch bottlenecks in mixture-of-experts parallelism across architectures.

Ax Islam Mansour, Ronny Haensch, Irena Hajnsek, Konstantinos Papathanassiou 5/21/2026

Hybrid Machine Learning Model for Forest Height Estimation from TanDEM-X and Landsat Data

Hybrid ML-physics model for forest height estimation from satellite interferometry data.

Ax Shubhada Agrawal, Siva Theja Maguluri, Martin Zubeldia 5/21/2026

Concentration of General Stochastic Approximation Under Heavy-Tailed Markovian Noise

Concentration bounds for stochastic approximation algorithms under heavy-tailed Markovian noise.

Ax Ishaan Kelkar, Nebras Alam, Vikram Kakaria, Madhur Panwar, Vasu Sharma, Maheep Chaudhary 5/21/2026

Playing Devil's Advocate: Off-the-Shelf Persona Vectors Rival Targeted Steering for Sycophancy

Study comparing persona steering vectors to targeted steering (CAA) for mitigating LLM sycophancy.

Ax Henry Moss, Lachlan Astfalck, Thomas Cowperthwaite, Colin Doumont, Sam Willis, Philipp Hennig, Christopher Nemeth, Andrew Zammit-Mangion 5/21/2026

Conditioning Gaussian Processes on Almost Anything

Research on conditioning Gaussian processes via diffusion models with closed-form guidance.

Ax Ondrej Galeta, Lukas Sekanina 5/21/2026

Genetic Programming with Transformer-Based Mutation for Approximate Circuit Design

Transformer-based mutation operator for Cartesian genetic programming in automated circuit design.

Ax Yan Li, Yunlong Deng, Yuewen Sun, Gongxu Luo, Kun Zhang, Guangyi Chen 5/21/2026

Multimodal LLMs under Pairwise Modalities

Training multimodal LLMs using paired modalities instead of fully aligned multi-way datasets.

Ax Nassim Ait Ali Braham, Aaron Banze, Conrad M. Albrecht, Julien Mairal, Jocelyn Chanussot, Xiao Xiang Zhu 5/21/2026

SpectralEarth-FM: Bringing Hyperspectral Imagery into Multimodal Earth Observation Pretraining

SpectralEarth-FM foundation model integrating hyperspectral imagery with multisensor Earth observation data.

Ax Shinnosuke Taksuka, Hideo Mukai 5/21/2026

Musical Attention Transformer: Music Generation Using a Music-Specific Attention Model

Musical Attention Transformer mechanism to improve music generation quality and reduce repetition.

Ax D. -M. Mei, K. Acharya, C. M. Adhikari, M. Adhikari, S. Aryal, B. V. Benson, K. Bhatta, S. Bhattarai, N. Budhathoki, A. M. Castillo, D. Chakraborty, S. Chhetri, S. Choudhury, T. A. Chowdhury, R. D. Cruz, B. Cui, S. Dhital, K. -M. Dong, R. Gapuz, A. Ghasemi, E. Z. Gnimpieba, B. D. S. Gurung, H. A. Hashim, R. I. Harry, K. -E. Hasin, M. K. Hassanzadeh, M. K. Jha, D. Kim, K. -C. Kong, B. Lama, A. Mahat, N. Maharjan, A. Majeed, J. Mammo, M. M. Masud, K. S. Moore, A. Nawaz, H. Oli, S. A. Panamaldeniya, L. Pandey, R. Pandey, Z. Peng, A. Prem, M. M. Rana, K. Rana Magar, R. Rizk, C. S. Tadi, L. -W. Wang, Y. Yang, G. -L. Yin, C. -X. Yu, D. Zeng, M. Zhou, Q. Zhou 5/21/2026

AIMBio-Mat: An AI-Native FAIR Platform for Closed-Loop Materials Discovery and Biomedical Translation

AIMBio-Mat framework for AI-guided materials discovery and biomedical translation with FAIR principles.

Ax Alexi Canesse, Beno\^it Goupil, Jesse Read, Sonia Vanier 5/21/2026

Decoupling Communication from Policy: Robust MARL under Bandwidth Constraints

Research on decoupling communication from policy in multi-agent RL under bandwidth constraints.

Ax Mouhamed Amine Bouchiha, Abdelaziz Amara Korba, Yacine Ghamri-Doudane 5/21/2026

Automated Byzantine-Resilient Clustered Decentralized Federated Learning for Battery Intelligence in Connected EVs

Byzantine-resilient decentralized federated learning framework (ABC-DFL) for EV battery intelligence.

Ax Kesong Li, Yixuan Xu, Kuo-kun Tseng, Weiyi Lu, Kan Liu, Tao Lan 5/21/2026

Linear-DPO: Linear Direct Preference Optimization for Diffusion and Flow-Matching Generative Models

Research proposes Linear-DPO for aligning diffusion and flow-matching models via preference optimization.

Ax Yang Wu, Qiang Meng, Zhaojiang Liu, Youquan Liu, Jian Yang, Jin Xie 5/21/2026

Distill to Think, Foresee to Act: Cognitive-Physical Reinforcement Learning for Autonomous Driving

End-to-end autonomous driving using reinforcement learning with cognitive and foresight components to overcome imitation learning limitations.

Ax Fernando Ortega, Ra\'ul Lara-Cabrera, Jorge Due\~nas-Ler\'in, Alejandro de la Torre-Luque, Merc\'e Salvador Robert, Enrique Baca-Garc\'ia 5/21/2026

Automated ICD Classification of Psychiatric Diagnoses: From Classical NLP to Large Language Models

Automates ICD psychiatric diagnosis classification from Spanish free-text descriptions using NLP and ML techniques on 145,513 clinical records.

Ax Oskar Allerbo, Thomas B. Sch\"on 5/21/2026

A Rigorous, Tractable Measure of Model Complexity

Proposes mathematically rigorous, computationally efficient model complexity measure based on gradient similarities across inputs for interpretation and model selection.

Ax Austin Braniff, Yuhe Tian 5/21/2026

Reinforcement Learning-based Control via Y-wise Affine Neural Networks: Comparative Case Studies for Chemical Processes

RL-based control approach using Y-wise affine neural networks for chemical process management.

Ax Austin Braniff (Department of Chemical and Biomedical Engineering, West Virginia University), Fengqi You (R.F. Smith School of Chemical and Biomolecular Engineering, Cornell University), Yuhe Tian (Department of Chemical and Biomedical Engineering, West Virginia University) 5/21/2026

Enhanced Reinforcement Learning-based Process Synthesis via Quantum Computing

Quantum-enhanced reinforcement learning framework for chemical process synthesis with improved scalability.

Ax Shuaida He, Liwen Chen, Long Feng 5/21/2026

Federated LoRA Fine-Tuning for LLMs via Collaborative Alignment

Federated learning framework for parameter-efficient LoRA fine-tuning of LLMs with heterogeneous clients.

Ax Krishnakumar Balasubramanian 5/21/2026

Large-Step Training Dynamics of a Two-Factor Linear Transformer Model

Gradient descent analysis of simplified linear transformer learning in-context regression at large learning rates.

Ax Yicheng Feng, Xin Tan, Yangtao Deng, Yimin Jiang, Yibo Zhu, Hong Xu 5/21/2026

Frontier: Towards Comprehensive and Accurate LLM Inference Simulation

High-fidelity LLM inference simulator supporting disaggregated execution, complex parallelism, and agent workloads.

Ax Lucheng Fu, Ye Yu, Yiyang Wang, Yiqiao Jin, Haibo Jin, B. Aditya Prakash, Haohan Wang 5/21/2026

TextReg: Mitigating Prompt Distributional Overfitting via Regularized Text-Space Optimization

Prompt optimization method using regularization to prevent distributional overfitting and improve LLM generalization.

Ax Fares El Khoury, Houssam Zenati, Nathan Kallus, Michael Arbel, Aur\'elien Bibaut 5/21/2026

Semiparametric Efficient Bilevel Gradient Estimation

Semiparametric debiasing theory for bilevel gradient estimation using efficient influence functions.

Ax Akshay Manglik (Emily), Apaar Shanker (Emily), Kaustubh Deshpande (Emily), Jason Qin (Emily), Yash Maurya (Emily), Veronica Chatrath (Emily), Vijay S. Kalmath (Emily), Levi Lentz (Emily), Yuan (Emily), Xue 5/21/2026

Insights Generator: Systematic Corpus-Level Trace Diagnostics for LLM Agents

Systematic corpus-level trace diagnostics tool for identifying and diagnosing failure patterns in LLM agent execution traces.

Ax Hongzhi Ruan, Pei Liu, Weiliang Ma, Zhengning Li, Xueyang Zhang, Jun Ma, Dan Xu, Kun Zhan 5/21/2026

Closed Loop Dynamic Driving Data Mixture for Real-Synthetic Co-Training

Data mixture optimization framework for efficient real-synthetic co-training in autonomous driving end-to-end learning.

Ax Yi Liu, Jia Ma, Wengen Li, Jihong Guan, Shuigeng Zhou, Yichao Zhang 5/21/2026

Disentangling Generation and Regression in Stochastic Interpolants for Controllable Image Restoration

Hybrid method combining generative and regression approaches for fast, efficient image restoration via stochastic interpolants.

Ax Antoine Maillard, Sebastian Goldt 5/21/2026

Memorisation, convergence and generalisation in generative models

Theoretical analysis of memorization vs. generalization in diffusion models through independent training on dataset subsets.

Ax Elle Miller, Jayaram Reddy, Ayush Deshmukh, Trevor McInroe, David Abel, Oisin Mac Aodha, Sethu Vijayakumar 5/21/2026

roto 2.0: The Robot Tactile Olympiad

Benchmark for standardizing tactile-based reinforcement learning across robotic morphologies with GPU parallelization.

Ax Harsh Parikh, Gabriel Levin-Konigsberg, Dominique Perrault-Joncas, Alexander Volfovsky 5/21/2026

Mind the Sim-to-Real Gap & Think Like a Scientist

Framework for deciding when to supplement pre-trained simulators with real experiments under budget constraints.

Ax Junshu Pan, Panzhong Lu, Yixuan Weng, Qiyao Sun, Fang Guo, Zijie Yang, Qiji Zhou, Yue Zhang 5/21/2026

AiraXiv: An AI-Driven Open-Access Platform for Human and AI Scientists

AI-driven platform for publishing and organizing human and AI-generated research, addressing scalability in academic publishing.

Ax Wenjia Wang, Xiaowei Zhang 5/21/2026

On the Suboptimality of GP-UCB under Polynomial Effective Optimism

Theoretical analysis of GP-UCB optimality for sequential optimization of black-box functions through effective optimism levels.

Ax Mohamad Fares El Hajj Chehade, Amrit Singh Bedi, Amy Zhang, Hao Zhu 5/21/2026

TRAM: Test-Time Risk Adaptation with Mixture of Agents

TRAM enables test-time adaptation of RL agents to new safety constraints by compositing a mixture of pre-trained risk-neutral policies without retraining.

Ax Xingyu Xie, Kuangyu Ding, Shuicheng Yan, Kim-Chuan Toh, Tianwen Wei 5/21/2026

Optimization Hyper-parameter Laws for Large Language Models

Presents Optimization Hyper-parameter Laws framework for deriving dynamic learning rate schedules and other optimization parameters during LLM training.

Ax Sanghyeon Lee, Sangjun Bae, Yisak Park, Seungyul Han 5/21/2026

Self-Improving Skill Learning for Robust Skill-based Meta-Reinforcement Learning

Proposes self-improving mechanism for skill-based meta-RL to reduce sensitivity to noisy offline demonstrations in long-horizon environments.

Ax Prasanna Mayilvahanan, Thadd\"aus Wiedemer, Sayak Mallick, Matthias Bethge, Wieland Brendel 5/21/2026

LLMs on the Line: Data Determines Loss-to-Loss Scaling Laws

Investigates factors influencing loss-to-loss scaling laws that relate pretraining and downstream task losses for LLM optimization and generalization.

Ax Yannis Montreuil, Axel Carlier, Lai Xing Ng, Wei Tsang Ooi 5/21/2026

Why Ask One When You Can Ask $k$? Learning-to-Defer to the Top-$k$ Experts

Extends Learning-to-Defer framework to allocate queries to top-k experts instead of single expert, unifying multiple prediction paradigms.

Ax Keisuke Kawano, Takuro Kutsuna, Naoki Hayashi, Yasushi Esaki, Hidenori Tanaka 5/21/2026

CT-OT Flow: Estimating Continuous-Time Dynamics from Discrete Temporal Snapshots

Proposes CT-OT Flow to estimate continuous-time dynamics from discrete temporal snapshots using optimal transport methods.

Ax Jinbo Wang, Mingze Wang, Jiaqi Zhang, Wei Wang, Peng Pei, Xunliang Cai, Weinan E, Lei Wu 5/21/2026

GradPower: Powering Gradients for Faster Language Model Pre-Training

Introduces GradPower, a lightweight gradient transformation technique using sign-power elementwise operations to accelerate language model pre-training.

Ax Zesen Wang, Lijuan Lan, Yonggang Li 5/21/2026

Time-Prompt: Integrated Heterogeneous Prompts for Unlocking LLMs in Time Series Forecasting

Uses heterogeneous prompting techniques to improve LLM performance on time series forecasting tasks compared to traditional deep learning methods.

Ax Jaebak Hwang, Sanghyeon Lee, Jeongmo Kim, Seungyul Han 5/21/2026

Strict Subgoal Execution: Reliable Long-Horizon Planning in Hierarchical Reinforcement Learning

Proposes Strict Subgoal Execution for hierarchical RL in long-horizon tasks, improving subgoal feasibility and high-level planning reliability.

Ax Chenqing Lin, Mostafa Hussien, Chengyao Yu, Bingyi Jing, Ruixing Ming, Kim Khoa Nguyen, Mohamed Cheriet 5/21/2026

FAIR-Pruner: A Flexible Framework for Automatic Layer-Wise Pruning via Tolerance of Difference

Presents FAIR-Pruner, a search-free framework for adaptive layer-wise structured pruning of neural networks using within-layer ranking signals.

Ax Menghua Jiang, Yuxia Lin, Baoliang Chen, Haifeng Hu, Yuncheng Jiang, Sijie Mai 5/21/2026

Disentangling Bias by Modeling Intra- and Inter-modal Causal Attention for Multimodal Sentiment Analysis

Addresses spurious correlations in multimodal sentiment analysis using causal attention mechanisms across text, audio, and visual modalities.