Ax Xiaohang Yu, Hejia Geng, Xinmeng Zeng, William Knottenbelt 5/5/2026

SUDP: Secret-Use Delegation Protocol for Agentic Systems

SUDP: delegation protocol preventing credential exposure and account compromise in agentic systems accessing user secrets and APIs.

Ax Wenshuo Zhao (May), Qi Zhu (May), Xingshan Zeng (May), Fei Mi (May), Lifeng Shang (May), Yi R. (May), Fung 5/5/2026

Entropy Centroids as Intrinsic Rewards for Test-Time Scaling

Entropy Centroids: intrinsic reward mechanism for test-time scaling of LLM inference without requiring external reward models.

Ax Boris Shigida, Boris Hanin, Andrey Gromov 5/5/2026

Learning Rate Transfer in Normalized Transformers

Learning rate transfer improvements for Normalized Transformers using alignment exponents for better hyperparameter scaling across model sizes.

Ax Halley Young, Nikolaj Bj\"orner 5/5/2026

Theory Under Construction: Orchestrating Language Models for Research Software Where the Specification Evolves

Methodology for orchestrating LLMs in research software development where specifications evolve, addressing hallucination accumulation failures.

Ax Chenxin Li, Zhengyang Tang, Mingxin Huang, Yunlong Lin, Shijue Huang, Shengyuan Liu, Bowen Ye, Rang Li, Lei Li, Benyou Wang, Yixuan Yuan 5/5/2026

Claw-Eval-Live: A Live Agent Benchmark for Evolving Real-World Workflows

Claw-Eval-Live: live benchmark for workflow agents with refreshable task signals, execution verification, and evolving real-world task evaluation.

Ax Brice Valentin Kok-Shun, Johnny Chan, Gabrielle Peko, David Sundaram 5/5/2026

Agentopic: A Generative AI Agent Workflow for Explainable Topic Modeling

Agentopic: multi-agent LLM workflow for explainable topic modeling with collaborative topic identification, validation, and hierarchical grouping.

Ax Hao Xiao 5/5/2026

From Euler to Dormand-Prince: ODE Solvers for Flow Matching Generative Models

ODE Solvers for Flow Matching: PyTorch implementations and benchmarks of Euler, RK4, and Dormand-Prince solvers for generative model sampling.

Ax Duaa Alim, Mogtaba Alim, Liam Chalcroft 5/5/2026

GAZE: Grounded Agentic Zero-shot Evaluation with Viewer-Level Tools and Literature Retrieval on Rare Brain MRI

GAZE: agentic framework enabling medical vision-language models to iteratively inspect images and retrieve literature via tool-calling and retrieval mechanisms.

Ax Guantian Zheng 5/5/2026

StyleShield: Exposing the Fragility of AIGC Detectors through Continuous Controllable Style Transfer

StyleShield: adversarial attack framework exposing vulnerabilities of AI-generated content detectors through controllable style transfer attacks.

Ax Seyed Mohammad Azimi-Abarghouyi, Mehdi Bennis, Leandros Tassiulas 5/5/2026

Hierarchical Federated Learning for Networked AI: From Communication Saving to Architecture-Aware Design

Hierarchical federated learning architecture for distributed AI systems optimizing communication and computation across multi-tier networks.

Ax Hada Melino Muhammad, Zechen Li, Flora Salim, Ahmed A. Metwally 5/5/2026

CGM-JEPA: Learning Consistent Continuous Glucose Monitor Representations via Predictive Self-Supervised Pretraining

CGM-JEPA self-supervised pretraining for consistent continuous glucose monitoring representations across multiple modalities.

Ax Anton Kolonin 5/5/2026

Interpretable experiential learning based on state history and global feedback

Interpretable reinforcement learning model using state history and global feedback, evaluated on Atari environments.

Ax Zhi-Quan Feng, Ying-Jia Lin, Hung-Yu Kao 5/5/2026

Learning in the Fisher Subspace: A Guided Initialization for LoRA Fine-Tuning

Fisher subspace initialization method for LoRA fine-tuning of LLMs, improving adaptation performance through informed subspace selection.

Ax Shashank Kapadia, Deep Naryan Mishra, Sujal Reddy Alugubelli, Haoan Wang, Saipraveen Vabbilisetty, Rishi Bhatia, Anupriya Sharma 5/5/2026

LEAP: Layer-wise Exit-Aware Pretraining for Efficient Transformer Inference

LEAP enables efficient transformer inference by addressing incompatibility between layer-wise distillation and convergence-based early exit mechanisms.

Ax Bruno Abrahao 5/5/2026

GEODE: Angle-Adaptive OOD Detection with Universal Scorer Compatibility

GEODE framework for out-of-distribution detection with universal scorer compatibility, improving on Outlier Exposure method.

Ax James Butterworth, Gevik Grigorian, Alejandro DiazDelaO 5/5/2026

Deep Variational Inference Symbolic Regression

Bayesian symbolic regression using deep variational inference to generate probability distributions over interpretable expressions quantifying uncertainty.

Ax Devdutt Subhasish, Henrik Hose, Sebastian Trimpe 5/5/2026

Learning to Race in Minutes: Infoprop Dyna on the Mini Wheelbot

RL approach using Infoprop Dyna for fast robot learning without physics simulators achieving sim-to-real transfer on wheeled robot.

Ax Yilei Chen, Sharut Gupta, Yannis Paschalidis, Ayush Sekhari, Aldo Pacchiano 5/5/2026

When Less is Enough: Efficient Inference via Collaborative Reasoning

DUET framework combining capable and lightweight models for collaborative two-stage inference reducing computational cost.

Ax Steven Tang, Xinze Xiong, Anna Hakhverdyan, Andrew Patterson, Jacob Adkins, Jiamin He, Esraa Elelimy, Parham Mohammad Panahi, Martha White, Adam White 5/5/2026

Forager: a lightweight testbed for continual learning with partial observability in RL

Continual reinforcement learning testbed (Forager) for studying plasticity and partial observability in non-stationary environments.

Ax Caleb Talley, Vedant Tibrewal, Seun Adekunle, Weiwen Dong, Xinyu Wu, Fariha Sheikh 5/5/2026

Multi-Perspective Transformers in ARC-AGI-2 Challenge

Approach using multi-perspective transformers and test-time training to solve ARC-AGI-2 visual reasoning puzzles achieving 96.1% training accuracy.

Ax Tam Nguyen, Tu Anh Nguyen, Sina Alemohammad, Richard G. Baraniuk 5/5/2026

Minimizing Collateral Damage in Activation Steering

Method for controlling LLM behavior through activation steering while minimizing unintended changes to non-target feature directions.

Ax Elon Litman, Gabe Guo 5/5/2026

A Theory of Generalization in Deep Learning

Non-asymptotic generalization theory for deep learning explaining how neural tangent kernels partition output space into signal and noise channels.

Ax Zheng-An Chen, Pengxiao Lin, Zhi-Qin John Xu, Tao Luo 5/5/2026

Focus and Dilution: The Multi-stage Learning Process of Attention

Analysis of attention learning dynamics in transformers identifying focus-dilution cycles during training via gradient-flow analysis.

Ax Hillary Mutisya, John Mugane 5/5/2026

Attention Sinks in Massively Multilingual Neural Machine Translation:Discovery, Analysis, and Mitigation

Identifies and analyzes attention sinks in multilingual neural machine translation where non-content tokens capture 83-91% of cross-attention in NLLB-200 model.

Ax Xiaorui Wang, Fanda Fan, Chenxi Wang, Yuxuan Yang, Rui Tang, Kuoyu Gao, Simiao Pang, Yuanfeng Shang, Zhipeng Liu, Wanling Gao, Lei Wang, Jianfeng Zhan 5/5/2026

CombinationTS: A Modular Framework for Understanding Time-Series Forecasting Models

Proposes CombinationTS modular framework for time-series forecasting that decomposes models into components to identify true performance drivers.

Ax Ruthwik Reddy Doodipala, Pankaj Pandey, Pratheek Eranki, Carolina Torres-Rojas, Manob Jyoti Saikia, Ranganatha Sitaram 5/5/2026

Rhamba: Region-Aware Hybrid Attention-Mamba Framework for Self-Supervised Learning in Resting-State fMRI

Introduces Rhamba framework combining region-aware masking with hybrid Attention-Mamba architectures for self-supervised pretraining on resting-state fMRI data.

Ax Ruiquan Huang, Donghao Li, Yingbin Liang, Jing Yang 5/5/2026

Breaking the Computational Barrier: Provably Efficient Actor-Critic for Low-Rank MDPs

Proposes computationally efficient actor-critic algorithm for reinforcement learning with low-rank MDPs, establishing hierarchy of function approximation approaches.

Ax Harsh Goel, Akhil Udathu, Susmija Jabireddy, Pradnesh Kalkar, Atharva Parulekar 5/5/2026

S^3-R1: Learning to Retrieve and Answer Step-by-Step with Synthetic Data

Introduces S³-R1 using synthetic data and RL post-training for models to perform step-by-step retrieval and reasoning for question-answering with tool-use.

Ax Wen-Da Wei, Han-Bin Fang, Yang-Di Liu, Jiang-Xin Shi, James Kwok, Yu-Feng Li 5/5/2026

Activation Compression in LLMs: Theoretical Analysis and Efficient Algorithm

Develops theoretical framework and efficient algorithm for activation compression in LLM training, reducing memory requirements during backpropagation.

Ax Dewei Bai, Hongxiang Peng, Yunyun Zeng, Ziyu Zhang, Hong Qu 5/5/2026

Congestion-Aware Dynamic Axonal Delay for Spiking Neural Networks

Proposes dynamic axonal delay mechanism for spiking neural networks using congestion-awareness to improve spike alignment in event-driven tasks.

Ax Xiaoyu Yang, En Yu, Jie Lu 5/5/2026

Autonomous Drift Learning in Data Streams: A Unified Perspective

Presents unified perspective on autonomous drift learning in non-stationary data streams beyond temporal concept drift in increasingly complex learning systems.

Ax Wang Jian, Zhou Jianbo, Xiong Yuhao, Liu Zhenxia, Luo Wen, Yuan LinWang, Yu ZhaoYuan 5/5/2026

GA-VisAgent: A Multi-Agent application for code generation and visualization in interactive learning

Proposes GA-VisAgent, a multi-agent application for code generation and visualization in Geometric Algebra learning, lowering adoption barrier through automated tooling.

Ax Chuang Liu, Zelin Yao, Xueqi Ma, Luzhi Wang, Mukun Chen, Pinghua Xu, Wenbin Hu 5/5/2026

GraphSculptor: Sculpting Pre-training Coreset for Graph Self-supervised Learning

Introduces GraphSculptor for constructing efficient pre-training coresets in graph self-supervised learning by exploiting dataset redundancy without additional training signals.

Ax Jikai Jin, Vasilis Syrgkanis 5/5/2026

The Partial Testimony of Logs: Evaluation of Language Model Generation under Confounded Model Choice

Analyzes confounding bias in offline language model evaluation from logs and proposes using randomized experiments to estimate true model performance differences.

Ax Yannik Schnitzer, Alessandro Abate, David Parker 5/5/2026

Robust Parameter Learning for Uncertain MDPs

Develops robust parameter learning method for uncertain MDPs that captures dependencies in transition probabilities from shared latent quantities.

Ax Ruiqi Xue, Lei Yuan, Kainuo Cheng, Jing-Wen Yang, Yang Yu 5/5/2026

Model-Based Proactive Cost Generation for Learning Safe Policies Offline with Limited Violation Data

Proposes model-based approach for learning safe offline policies with limited violation data in constraint-satisfying scenarios without risky online interaction.

Ax Fang Yuan, Quanjun Yin, Siqi Shen, Yuxiang Xie, Junqiang Yang, Long Qin, Junjie Zeng, Qinglun Li 5/5/2026

PACE: Parameter Change for Unsupervised Environment Design

Introduces PACE method for unsupervised environment design in reinforcement learning using parameter change as reliable environment evaluation signal.

Ax Junhyeong Lee, Sangjin Jin, Yongjae Lee 5/5/2026

Decision-Focused Learning via Tangent-Space Projection of Prediction Error

Develops closed-form geometric characterization of regret gradients for Decision-Focused Learning, improving computational efficiency over existing surrogate loss methods.

Ax Ting-Yu Dai, Kingsley Nweye, Dev Niyogi, Zoltan Nagy 5/5/2026

Toward a foundational thermal model for residential buildings

Proposes physics-informed transformer architecture for foundational thermal modeling of buildings generalizable across diverse structures without building-specific calibration.

Ax Maniru Ibrahim 5/5/2026

Sequential Learning and Catastrophic Forgetting in Differentiable Resistor Networks

Studies sequential learning and catastrophic forgetting in differentiable resistor networks governed by physical constraints, analyzing gradient-based parameter adjustment.

Ax Yuxuan Xiao, Shengzhong Zhang 5/5/2026

Rethinking Multi-Label Node Classification: Do Tuned Classic GNNs Suffice?

Investigates whether carefully tuned baseline GNNs can match specialized methods for multi-label node classification, questioning the necessity of complex label-aware designs.

Ax Richeng Zhou, Xuelin Zhang, Liyuan Liu 5/5/2026

Quantifying Multimodal Capabilities: Formal Generalization Guarantees in Pairwise Metric Learning

Theoretical generalization analysis for multimodal metric learning addressing how modality selection affects performance under incomplete data.

Ax Aloni Cohen, Chenhao Zhang 5/5/2026

Barriers to Counterfactual Credit Attribution for Autoregressive Models

Theoretical analysis of counterfactual credit attribution for autoregressive generative models, formalizing credit to training data dependencies.

Ax Sunil Kumar Maurya, Xin Liu 5/5/2026

Evaluating LLMs on Large-Scale Graph Property Estimation via Random Walks

Benchmark evaluating LLM capabilities on graph property estimation tasks using random walk representations to test reasoning on structured data.

Ax Zixi Shao, Tao Wang, Yibei Xiao, Tianyi Huang 5/5/2026

Protein-Conditioned Multi-Objective Reinforcement Learning for Full-Length mRNA Design

ProMORNA uses multi-objective RL and encoder-decoder models to generate full-length mRNA sequences from protein targets, balancing stability and translation.

Ax Paul Garnier, Vincent Lannelongue, Elie Hachem 5/5/2026

Mesh Based Simulations with Spatial and Temporal awareness

Graph neural networks and transformers as ML surrogates for CFD simulations with improved training paradigms beyond node-wise supervision.

Ax T. Satyanarayana Murthy, B. Swathi Sowmya, Santhosh Voruganti, Sai Varshini Giridi, Chaitanyya Pratap Agarwal, Vanteddu Akshitha 5/5/2026

Hybrid Quantum Reinforcement Learning with QAOA for Improved Vehicle Routing Optimization

Hybrid quantum-classical reinforcement learning approach using QAOA for vehicle routing optimization, a combinatorial NP-hard problem.

Ax Donato Crisostomi 5/5/2026

Model Merging: Foundations and Algorithms

Thesis on model merging paradigm for combining independently trained neural networks in weight space without optimization or original training data.

Ax Pratyush Acharya, Nuraj Rimal, Habish Dhakal 5/5/2026

Concepts Whisper While Syntax Shouts: Spectral Anti-Concentration and the Dual Geometry of Transformer Representations

Analysis of transformer representations across languages, testing whether spectral properties enable cross-lingual concept transport.

Ax Daniel Agyei Asante, Ernie Chang, Yang Li 5/5/2026

Importance-Guided Basis Selection for Low-Rank Decomposition of Large Language Models

Method for selecting singular-vector bases in low-rank LLM decomposition using importance-guided approach aligned with loss landscape geometry.