Ax Xiaoguang Guo, Zehong Wang, Ziming Li, Shawn Spitzel, Soonwoo Kwon, Tianyi Ma, Yanfang Ye, Chuxu Zhang 5/8/2026

On the Safety of Graph Representation Learning

arXiv paper on safety and robustness of graph representation learning under distribution shifts. ML robustness research.

Ax Adhiraj Banerjee, Vipul Arora 5/8/2026

PairAlign: A Framework for Sequence Tokenization via Self-Alignment with Applications to Audio Tokenization

arXiv paper on discrete tokenization framework for audio via self-alignment. Research on audio representation learning.

Ax Vinit Ranjan, Jisun Park, Bartolomeo Stellato 5/8/2026

Distributionally-Robust Learning to Optimize

arXiv paper on distributionally robust hyperparameter optimization for convex methods. Theoretical optimization research.

Ax Abhijit Das, Sayantan Dutta 5/8/2026

Weight-Decay Turns Transformer Loss Landscapes Villani: Functional-Analytic Foundations for Optimization and Generalization

arXiv paper on weight decay effects on transformer loss landscapes using functional analysis. Theoretical optimization research for LLMs.

Ax Shai Feldman, Yaniv Romano 5/8/2026

How Many Iterations to Jailbreak? Dynamic Budget Allocation for Multi-Turn LLM Evaluation

arXiv paper on efficient multi-turn LLM evaluation and jailbreak prediction with adaptive budgets. Research on LLM robustness evaluation.

Ax Jakub St\k{e}pie\'n, Marcin Mazur, Jacek Tabor, Przemys{\l}aw Spurek 5/8/2026

SoftSAE: Dynamic Top-K Selection for Adaptive Sparse Autoencoders

arXiv paper on sparse autoencoders with dynamic sparsity selection for LLM interpretability. Research on mechanistic interpretability tools.

Ax Siquan Li, Kaiqi Jiang, Jiacheng Sun, Tianyang Hu 5/8/2026

The Structural Origin of Attention Sink: Variance Discrepancy, Super Neurons, and Dimension Disparity

arXiv paper analyzing mechanistic origins of attention sink phenomenon in LLMs. Original research on transformer internals.

Ax Yang Xu, Chiwoo Park 5/8/2026

Online Bayesian Calibration under Gradual and Abrupt System Changes

arXiv paper on Bayesian calibration for digital twins under system changes. Machine learning research with theoretical contribution.

Ax Hongyi Tao, Dingzhi Yu, Lijun Zhang 5/8/2026

When and Why SignSGD Outperforms SGD: A Theoretical Study Based on $\ell_1$-norm Lower Bounds

Theoretical analysis of why sign-based optimization algorithms outperform SGD in training foundation models.

Ax Yuping Lin, Pengfei He, Yue Xing, Yingqian Cui, Jiayuan Ding, Subhabrata Mukherjee, Hui Liu, Zhen Xiang 5/8/2026

Crafting Reversible SFT Behaviors in Large Language Models

Method for creating reversible SFT behaviors in LLMs with causal control over induced behaviors.

Ax Apurva Gandhi, Satyaki Chakraborty, Xiangjun Wang, Aviral Kumar, Graham Neubig 5/8/2026

Recursive Agent Optimization

Reinforcement learning approach for training recursive agents that spawn sub-tasks to solve longer context problems.

Ax Ronaldo Canizales, Divya Gopinath, Corina P\u{a}s\u{a}reanu, Ravi Mangal 5/8/2026

Concept-Based Abductive and Contrastive Explanations for Behaviors of Vision Models

Concept-based explanation method for vision models using causal abductive and contrastive explanations.

Ax Sushant Gautam, Finn Schwall, Annika Willoch Olstad, Fernando Vallecillos Ruiz, Birk Torpmann-Hagen, Sunniva Maria Stordal Bj{\o}rklund, Leon Moonen, Klas Pettersen, Michael A. Riegler 5/8/2026

When No Benchmark Exists: Validating Comparative LLM Safety Scoring Without Ground-Truth Labels

Framework for validating LLM safety scores without labeled benchmarks using scenario-based audits.

Ax Yuxing Liu, Jianyu Wang, Tong Zhang 5/8/2026

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less

Study showing using same optimizer in finetuning as pretraining reduces catastrophic forgetting in LLMs.

Ax Jai Moondra, Ayela Chughtai, Bhargavi Lanka, Swati Gupta 5/8/2026

Why Global LLM Leaderboards Are Misleading: Small Portfolios for Heterogeneous Supervised ML

Analysis showing global LLM leaderboards are statistically misleading due to large vote cancellation across languages.

Ax Yuhang Lai, Jiazhan Feng, Yee Whye Teh, Ning Miao 5/8/2026

Verifier-Backed Hard Problem Generation for Mathematical Reasoning

Method for generating hard mathematical problems using LLM verifiers for improved training datasets.

Ax Olivia Zhang, Zhilin Zhang 5/8/2026

A Review of Large Language Models for Stock Price Forecasting from a Hedge-Fund Perspective

Review of LLM applications in quantitative finance including sentiment analysis, multi-agent trading systems, and practical pitfalls.

Ax Guoqing Cai, Kai Zeng, Shoulin Huang, Ting Ma 5/8/2026

MPNet: A Robust and Efficient Manifold Pooling Network for Multi-Rhythm EEG Signal Decoding

Manifold pooling network for efficient EEG signal decoding with Riemannian geometry approaches.

Ax ZhengXiao He, Huayu Li, Xiwen Chen, Janet M Roveda, Jinghao Wen, Siyuan Tian, Ao Li 5/8/2026

MedMamba: Recasting Mamba for Medical Time Series Classification

Mamba-based architecture adapted for medical time series classification tasks like ECG and EEG analysis.

Ax Jinxing Li, Nicholas Ren, Cathy Chang, Hongkai Pan, Daniel George 5/8/2026

Layout-Aware Representation Learning for Open-Set ID Fraud Discovery

Open-set fraud detection for identity documents using layout-aware representation learning and adapted vision models.

Ax Aadarsh Senapati, Neha Kujur, Vivek Yelleti 5/8/2026

Dynamic Graph with Similarity-Aware Attention Graph Neural Network for Recommender Systems

Graph neural network approach for recommender systems using dynamic graphs and similarity-aware attention.

Ax Eric Wolos, Michael Doyle 5/8/2026

Identifier-Free Code Embedding Models for Scalable Search

Embedding models for bidirectional code search between source and decompiled code without identifiers.

Ax Yuelin Hu, Zhenbo Yu, Zhengxue Cheng, Wei Liu, Li Song 5/8/2026

Maximizing Rollout Informativeness under a Fixed Budget: A Submodular View of Tree Search for Tool-Use Agentic Reinforcement Learning

Tree search optimization for tool-use agent RL via submodular maximization of rollout informativeness in GRPO.

Ax Daniel Grimmer 5/8/2026

Direct From Darwin: Deriving Advanced Optimizers From Evolutionary First Principles

Advanced gradient-based optimizers derived from evolutionary computation first principles.

Ax Alex Oesterling, Donghao Ren, Yannick Assogba, Dominik Moritz, Sunnie S. Y. Kim, Leon Gatys, Fred Hohman 5/8/2026

Understanding Annotator Safety Policy with Interpretability

Interpretability analysis of annotator safety policies identifying sources of disagreement in AI safety annotation.

Ax Philippe Hansen-Estruch, Jiahui Chen, Vivek Ramanujan, Orr Zohar, Yan Ping, Animesh Sinha, Markos Georgopoulos, Edgar Schoenfeld, Ji Hou, Felix Juefei-Xu, Sriram Vishwanath, Ali Thabet 5/8/2026

ViTok-v2: Scaling Native Resolution Auto-Encoders to 5 Billion Parameters

ViTok-v2: Scaled Vision Transformer autoencoders to 5B parameters for image tokenization at native resolution.

Ax Becky Langdon, Gabriel D. Patr\'on, Chrysoula D. Kappatou, Robert M. Lee, Behrang Shafei, Jixiang Qing, Ruth Misener, Mark van der Wilk, Calvin Tsay 5/8/2026

Meta-learning for sample-efficient Bayesian optimisation of fed-batch processes

Meta-learning approach for sample-efficient Bayesian optimization of fed-batch chemical processes.

Ax Aymen Echarghaoui, Dongxia Wu, Emily B. Fox 5/8/2026

BALAR : A Bayesian Agentic Loop for Active Reasoning

BALAR: Bayesian agentic loop algorithm for LLMs to actively determine missing information and ask next questions.

Ax Joseph H. Rudoler, Kevin Tan, Giles Hooker, Konrad P. Kording 5/8/2026

Estimating Implicit Regularization in Deep Learning

Method for estimating implicit regularization in deep learning models without analytical derivation.

Ax Kennedy Edemacu, Mohammad Mahdi Shokri, Vinay M. Shashidhar, Jong Wook Kim 5/8/2026

Privacy Without Losing Place: A Paradigm for Private Retrieval in Spatial RAGs

PAS mechanism for location privacy in spatial retrieval-augmented generation systems using anchor-based encoding.

Ax Langlin Huang, Chengsong Huang, Jinyuan Li, Donghong Cai, Yuyi Yang, Jiaxin Huang 5/8/2026

Nonsense Helps: Prompt Space Perturbation Broadens Reasoning Exploration

LLM reasoning improvement via prompt space perturbation to address zero-advantage problem in Group Relative Policy Optimization training.

Ax Siyan Liu, Yi Chang, Manli Cheng, Qinglong Tian, Pengfei Li 5/8/2026

In-Context Positive-Unlabeled Learning

Pretrained transformer (PUICL) for in-context positive-unlabeled learning enabling quick binary classification with only positive labels available.

Ax Hoin Jung, Xiaoqian Wang 5/8/2026

The Cost of Context: Mitigating Textual Bias in Multimodal Retrieval-Augmented Generation

Analysis of recorruption failure mode in multimodal RAG systems where accurate retrieved context causes MLLMs to abandon correct predictions.

Ax Yu Wang, Arnab Ganguly 5/8/2026

Variational Smoothing and Inference for SDEs from Sparse Data with Dynamic Neural Flows

Variational inference method for calibrating stochastic differential equations from sparse observations using dynamic neural flows.

Ax Gabriel Jeanson, David-Alexandre Duclos, William Larriv\'ee-Hardy, No\'e Cochet, Mat\v{e}j Boxan, Anthony Desch\^enes, Fran\c{c}ois Pomerleau, Philippe Gigu\`ere 5/8/2026

Leveraging Image Generators to Address Training Data Scarcity: The Gen4Regen Dataset for Forest Regeneration Mapping

Synthetic dataset (Gen4Regen) using image generators to address training data scarcity for forest regeneration mapping with deep learning.

Ax Jannis Chemseddine, Gregor Kornhardt, Gabriele Steidl 5/8/2026

Spherical Flows for Sampling Categorical Data

Generative model for discrete sequences using spherical flows and von Mises-Fisher distributions with closed-form conditional scores.

Ax Samuel Korn 5/8/2026

Architecture Matters: Comparing RAG Systems under Knowledge Base Poisoning

Evaluation of RAG system architectures including multi-agent debate and agentic retrieval under adversarial knowledge base poisoning attacks.

Ax Dongfang Zhao 5/8/2026

EGA: Adapting Frozen Encoders for Vector Search with Bounded Out-of-Distribution Degradation

Residual adapter method (EGA) for frozen vision encoders in vector search with robustness to out-of-distribution queries.

Ax Andy Zeyi Liu, Elliot Paquette, John Sous 5/8/2026

Spectral Lens: Activation and Gradient Spectra as Diagnostics of LLM Optimization

Spectral analysis framework using activation and gradient covariance to diagnose internal mechanisms during language model training.

Ax Yanlong Zhao, Xiaoyuan Cheng, Huihang Liu, Baihua He, Xinyu Zhang, Harrison Bo Hua Zhu, Wenlong Chen, Li Zeng, Zhuo Sun 5/8/2026

Saliency-Aware Regularized Quantization Calibration for Large Language Models

Saliency-aware quantization calibration method for post-training quantization of large language models with improved generalization.

Ax Bole Ma, Jan Eitzinger, Harald K\"ostler 5/8/2026

Irminsul: MLA-Native Position-Independent Caching for Agentic LLM Serving

Position-independent caching system (Irminsul) for agentic LLM serving that prevents cache invalidation from token shifts during multi-turn interactions.

Ax Huchen Yang, Xinghao Dong, Dan Negrut, Jin-Long Wu 5/8/2026

Active Learning for Communication Structure Optimization in LLM-Based Multi-Agent Systems

Active learning method for optimizing communication structures in LLM multi-agent systems to reduce token usage and improve performance.

Ax Anh H. Vo, Sungyo Lee, Phil-Joong Kim, Soo-Mi Choi, Yong-Guk Kim 5/8/2026

Closing the Loop: Unified 3D Scene Generation and Immersive Interaction via LLM-RL Coupling

Framework combining LLMs with reinforcement learning for unified 3D scene generation and interactive user interaction in multimedia systems.

Ax Ming Liu 5/8/2026

Decodable but Not Corrected by Fixed Residual-Stream Linear Steering: Evidence from Medical LLM Failure Regimes

Research on linear decodability versus correction of medical LLM failure modes, showing limitations of fixed residual-stream steering.

Ax Joseph D. Ramsey 5/8/2026

Fourier Feature Methods for Nonlinear Causal Discovery: FFML Scoring and FFCI Testing in Mixed Data

Research on Fourier feature methods for nonlinear causal discovery in mixed data combining scoring and constraint-based approaches.

Ax Dongjin Kim, Daniel S. King, Yoonjae Park, Roya Savoj, Sebastien Hamel, Xiaoyu Wang, Bingqing Cheng 5/8/2026

Polarizable atomic multipoles for learning long-range electrostatics

Research on machine learning interatomic potentials with polarizable atomic multipoles for modeling long-range electrostatics.

Ax Haodong Liang, Lifeng Lai 5/8/2026

Transformers Provably Implement In-Context Reinforcement Learning with Policy Improvement

Theoretical work proving transformers can implement in-context reinforcement learning with policy improvement via explicit constructions.

Ax Yuqian Cheng, Zhuo Chen, Qian Lin 5/8/2026

Optimal Confidence Band for Kernel Gradient Flow Estimator

Research on supremum-norm generalization error and uniform inference bounds for kernel gradient flow methods.

Ax Lena Helgerth, Andreas Christmann 5/8/2026

Ratio-based Loss Functions

Survey of ratio-based loss functions for supervised and unsupervised learning algorithms.

Ax Hirofumi Ota, Naoto Iwase, Yuki Ichihara, Junpei Komiyama, Masaaki Imaizumi 5/8/2026

CITE: Anytime-Valid Statistical Inference in LLM Self-Consistency

Research on anytime-valid statistical inference for controlling error in LLM self-consistency aggregation methods.