Ax Zhiyuan Chen, Jing Hu, Junzhe Wang, Yueyang Huang, Xinyi Yang, Zhaoyang Wang, Feng Zhu 7/8/2026

AbICL: In-Context Learning for Antigen-Specific Antibody Affinity Ranking

In-context learning applied to antibody affinity ranking for drug discovery by capturing antigen-specific binding landscapes.

Ax Shuze Daniel Liu, Claire Chen, Jiabao Sean Xiao, Xin Chen, David Simchi-Levi 7/8/2026

Strategic Bargaining in Multi-Buyer Markets: Reinforcement Learning from Verifiable Rewards for LLM Negotiations

arXiv paper on multi-buyer negotiation using reinforcement learning with LLMs for strategic bargaining with private information.

Ax Pan Li, Kai Chen, Shuai Chang, Shengzhi Zhang, Peizhuo Lv, Jinwen He 7/8/2026

Differentially Private Natural Gradient Descent

arXiv paper on differentially private natural gradient descent improving optimization efficiency under privacy constraints.

Ax Noel Thomas 7/8/2026

No Subspace to Track: Non-Identifiability and Optimizer State in Low-Rank Training

arXiv paper demonstrating non-identifiability of subspaces in low-rank LLM training, challenging GaLore optimizer assumptions.

Ax Sahasrajit Sarmasarkar, Anastasia Koloskova, Sanmi Koyejo 7/8/2026

Auditing of Unlearning Algorithms

arXiv paper proposing practical auditing of unlearning algorithms using membership inference attacks to verify data removal.

Ax Jean-Francois Bonbhel 7/8/2026

K-ABENA: K-Adaptive Backpropagation with Error-based N-exclusion Algorithm : (Compensated Loss-Based Sample Exclusion with Unbiased Gradient Estimation)

arXiv paper introducing K-ABENA, a selective gradient computation framework reducing training costs via sample exclusion and unbiased estimation.

Ax Chenyu Zhou 7/8/2026

More Convincing, Not More Correct: Self-Play Reward Hacking of Reference-Free LLM Judges

Research paper showing LLM judges optimize for plausibility over correctness in self-play settings, enabling reward hacking.

Ax Robin Holzinger (Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, USA), Riccardo Colletti (Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, USA) 7/8/2026

Drift Happens: An Empirical Study of Neural Architecture Robustness to Temporal Distribution Shift

arXiv empirical study of neural architecture robustness to temporal distribution shift across time-indexed domains.

Ax Ming Cai, Hisayuki Hara 7/8/2026

Learning Sparsest Linear Causal DAGs with Latent Confounders via Higher-Order Cumulants

arXiv paper on recovering sparse linear causal DAGs with latent confounders using higher-order cumulants.

Ax Xiangwu Wang, Chengwei Cao, Yicheng Song, Ran Bi, Peilin Yu 7/8/2026

Stability Annealing Selects the Implicit Bias of Smoothed Sign Descent: A Rate-Indexed Barrier Path on Separable Data

arXiv paper on stability annealing in smoothed sign descent; theoretical ML research on convergence and implicit bias.

Ax Giovanni Montanari, Marco Scarsini, Vianney Perchet 7/8/2026

Learning When to Automate: Queue Control in Human-AI Service Systems

arXiv paper on learning resource allocation between automated chatbots and human agents in sequential task processing systems.

Ax Shizhou Luo, Xiaodong Wei 7/8/2026

SplineNet: An Isogeometric Deep Learning Method for Complex Shells

arXiv paper on SplineNet, an isogeometric deep learning method for designing complex shell structures using B-spline representations.

Ax Taiki Yamada, Kantaro Fujiwara 7/8/2026

Scalable Perturbation Learning for Online Self-Supervised Echo State Networks

arXiv paper on online self-supervised learning for echo state networks with perturbation-based methods for autonomous adaptation.

Ax Alexander Apartsin, Yehudit Aperstein 7/8/2026

Modeling Normal Is All You Need: Joint Latent Clustering for Anomaly Detection in Multimodal Cyber-Physical Systems

arXiv paper on anomaly detection in cyber-physical systems using joint latent clustering to model normal behavior rather than faults.

Ax Xin Peng, Ang Gao 7/8/2026

x-Prediction Is All You Need:Training-Free Accelerated Generation via Endpoint Decodability

Training-free method for accelerating diffusion and flow matching model sampling using x-prediction without retraining or distillation.

Ax Yao Fu, Chunxia Zhang, Junmin Liu, Yihang Jin, Haishan Ye, Yuanao Yang 7/8/2026

Leveraging Extragradient for Effective Sharpness-Aware Minimization in Deep Learning

Novel optimizer combining extragradient methods with sharpness-aware minimization to improve generalization in deep learning by finding flatter minima.

Ax Jie Huang, Pengfei Yin, Zihan Xu, Daniel Capurro, Mike Conway, Ting Dang 7/8/2026

X-FEMR: A Token-level Explainable Approach for Electronic Health Records Foundation Models using Transformer-based Models

Token-level explainability method for electronic health record foundation models using transformer attention analysis for clinical interpretability.

Ax Andrea Agazzi, Eloy Mosig Garc\'ia, Dario Trevisan 7/8/2026

Quantitative Gaussian-Process limits of Tensor Programs

Theoretical analysis of infinite-width Gaussian-process limits for random neural networks using tensor programs with quantitative convergence bounds.

Ax Raul Jimenez, Svitlana Mayboroda, Pavlos Protopapas, Leonid Sarieddine, David N. Spergel, Pedro Taranc\'on-\'Alvarez 7/8/2026

Physics-Informed Neural Embeddings of PDE Solution Families

Physics-informed neural network framework for learning embeddings of PDE solution families using multihead architecture.

Ax Naveen George, Naoki Murata, Yuhta Takida, Konda Reddy Mopuri, Yuki Mitsufuji 7/8/2026

TILDE: TILt-based Distributional Erasure for Concept Unlearning

TILDE: Novel machine learning method for concept unlearning in text-to-image diffusion models while preserving model quality.

Ax Przemys{\l}aw Rola 7/8/2026

EntroPath: Maximum Entropy Path Ensemble Embedding for Manifold Learning

EntroPath method for manifold learning using maximum entropy diffusion path ensembles. ArXiv research on geometric recovery from graphs.

Ax Shervin Khalafi, Igor Krawczuk, Sergio Rozada, Charilaos Kanatsoulis, Antonio G Marques, Alejandro Ribeiro 7/8/2026

Graph Convolutional Attention: A Spectral Perspective on Graph Denoising and Diffusion

Spectral analysis of attention mechanisms for graph denoising showing linear attention alignment with denoising objectives.

Ax Juntong Shi, Brian L. Trippe, Jure Leskovec, Stefano Ermon, Minkai Xu 7/8/2026

Diffusion Language Model Parallel Decoding via Product-of-Experts Bridge

Product-of-experts bridge for parallel decoding in diffusion language models improving generation quality over standard importance sampling.

Ax Chan Li, Nigel Goldenfeld 7/8/2026

Broken Ergodicity and the Violation of the Fluctuation-Dissipation Theorem Lead to Generalization Beyond Overfitting in Machine Learning

Dynamical mean field theory explaining generalization in overparameterized neural networks via broken ergodicity and fluctuation-dissipation violation.

Ax Amitash Nanda, Javier Hernandez Nicolau, Madhusudan Gujral, Mahidhar Tatineni, Amitava Majumdar, Debashis Sahoo 7/8/2026

Performance Optimization and Comparative Analysis of Generative AI Models on Advanced Accelerators

Performance optimization and comparative analysis of generative AI models across heterogeneous accelerators for deployment efficiency.

Ax Marc L\'eobet, Pierre-Fran\c{c}ois Lavall\'ee, Jean-Pierre Lorr\'e 7/8/2026

Life Cycle Assessment of Pre-training the Lucie 7B Open-Source Large Language Model on the Jean Zay Supercomputer

Life cycle assessment of Lucie 7B LLM pre-training covering operational, embodied emissions, and water consumption on HPC infrastructure.

Ax Paul K. Mandal, Pavan Reddy, Tristan Malatynski 7/8/2026

Statistical Adversaries: Natural Backdoor-like Features in Vision Datasets

Analysis of naturally occurring statistical patterns in ImageNet that function like backdoor triggers without malicious insertion.

Ax Giulia Lanzillotta, Mandana Samiei, Doina Precup, Razvan Pascanu, Claire Vernade 7/8/2026

To Retain or to Adapt? Generalizing Continual Learning

Challenges retention-centered continual learning paradigm, proposing adaptation-focused approach for non-stationary environments.

Ax Leonardo Trentini, Fanny Lehmann, Laura Crocetti, Benedikt Soja 7/8/2026

Integrating GNSS-Derived Zenith Wet Delay into a Weather Foundation Model Improves Precipitation Forecasting

REVIVE multi-modal framework for detecting and recovering autonomous vehicle cameras from vandalism-induced occlusion attacks.

Ax Abinav Kalyanasundaram, Karthikeyan Chandra Sekaran, Wolfgang Utschick, Michael Botsch 7/8/2026

Physics-Regularized Machine Learning for Proprioceptive Vehicle Localization Using Onboard Sensors

Foundation model enhanced with GNSS-derived atmospheric water vapor data for improved precipitation forecasting.

Ax Abinav Kalyanasundaram, Karthikeyan Chandra Sekaran, Wolfgang Utschick, Michael Botsch 7/8/2026

Uncertainty-Aware Velocity Correction for Proprioceptive Vehicle Localization using Evidential Mamba

ML approach using physics-based regularization for IMU-based vehicle localization without GNSS.

Ax Hunter Heidenreich 7/8/2026

Where to cut, how deep: BPE and Unigram-LM on chemistry SMILES

EVC-Mamba method for GNSS-denied vehicle localization using evidential deep learning to correct inertial drift.

Ax Xiaopu Wang, Zelin He, Chengyuan Liu, Runze Li 7/8/2026

Beyond Heuristic Tuning: Power-Calibrated LLM Watermarking

Controlled study comparing BPE and Unigram-LM tokenizers for chemistry SMILES representations in chemical language models.

Ax Amy Lu, Changxiu Ji 7/8/2026

Association Restoration Test: Revealing Restorable Shortcuts after Unlearning

Association Restoration Test diagnostic for evaluating whether unlearned label-attribute shortcuts remain functionally usable by classifiers.

Ax Yu Liu, Boris Slautin, Ian Mercer, Jon-Paul Maria, Sergei V. Kalinin 7/8/2026

From Closed-Loop Optimization to Open Decision Making: Coupled Digital Twins for Predictive and Autonomous Microscopy

Coupled digital-twin framework for autonomous microscopy combining predictive models of sample response and instrument detection.

Ax Praneeth Narisetty, Uday Kumar Reddy Kattamanchi, Shiva Nagendra Babu Kore 7/8/2026

Onnes: A Physics-Grounded Multi-Agent LLM Simulator for Cryogenic Fault Diagnosis in Quantum Computing Infrastructure

Onnes: multi-agent LLM simulator with physics-grounded digital twin for cryogenic fault diagnosis in quantum computing infrastructure.

Ax Kaishen Wang, Tong Zheng, Xuehao Cui, Ruibo Chen, Tianyi Xiong, Heng Huang 7/8/2026

Mitigating Factual Hallucination in Large Reasoning Models via Mixed-Mode Advantage Regularization

Mixed-mode advantage regularization technique to mitigate factual hallucinations in large reasoning models during question answering tasks.

Ax Cemil-Andrei Dilmac, Florinel-Alin Croitoru, Radu Tudor Ionescu 7/8/2026

Few-Medoids: An Embarrassingly Simple Coreset Selection Method for Few-Shot Knowledge Distillation

Few-medoids coreset selection method for efficient few-shot knowledge distillation using simple representative subset identification.

Ax Jurn-Gyu Park, Sanzhar Zholdybayev, Aidar Amangeldi, Ademi Zhanuzakova 7/8/2026

Energy-Efficient GPU DVFS for Fine-Tuning of SLMs on Resource-constrained Embedded Devices

Dynamic Voltage Frequency Scaling methods for energy-efficient fine-tuning of small language models on embedded GPU devices.

Ax Daniel Maninger, Leon Chemnitz, Jannis Brugger, Tushar Lamba, Amir Molzam Sharifloo, Mira Mezini 7/8/2026

Mitigating Errors in LLM-Generated Web API Invocations via Retrieval-Augmented Generation and Constrained Decoding

Techniques using retrieval-augmented generation and constrained decoding to improve LLM accuracy in generating correct web API invocation code.

Ax Phat Tran, Artin Lahni, Pranav Kulkarni, Yaolun Zhang 7/8/2026

Is Domain Adaptation Always Helpful? A Frozen-Backbone Study of Cross-Domain Sentiment Transfer

Case study evaluating domain adaptation benefits for sentiment analysis with frozen pre-trained language model backbones of varying sizes.

Ax Soohyeon Choi, Debin Gao, Yue Duan 7/8/2026

Multi-Channel Spread-Spectrum Code Watermarking

Multi-channel spread-spectrum watermarking scheme for LLM-generated code to enable attribution and provenance tracking with high payload capacity.

Ax Alexander Rombach, Chantale Lauer, Nijat Mehdiyev 7/8/2026

Improving LLM-Generated Process Model Quality Through Reinforcement Learning: The Role of Reward Function Design

Systematic study of reinforcement learning reward function design for improving LLM-generated BPMN process models.

Ax Hong Lyu, Mingru Yang, Qianhua He, Yanxiong Li, Jinxin Huang, Zhengyu Pei 7/8/2026

TriA Pipeline: A Large-Scale Automatic Audio Annotation Pipeline For Audio Classification In Specific Scenarios

Automatic audio annotation pipeline for converting unlabeled domestic audio into labeled training data for classification.

Ax Ryuji Oi, Hikari Otsuka, Kosuke Matsushima, Yuki Ichikawa, Masato Motomura, Tatsuya Kaneko, Daichi Fujiki 7/8/2026

Training-Free Acceleration for Vision-Language-Action Models with Action Caching and Refinement

Training-free acceleration technique for Vision-Language-Action models using action caching for faster robotic inference.

Ax Arkaprabha Ganguli, Emil Constantinescu 7/8/2026

A Function-Space Dichotomy for Compositional Learning: Exponential Sub-Optimality of the Neural Tangent Kernel

Theoretical analysis of neural networks outperforming neural tangent kernel on compositional tasks, quantifying performance gaps.

Ax Tianjiao Yu, Xinzhuo Li, Yifan Shen, Onkar Susladkar, Yuanzhe Liu, Xiaona Zhou, Ismini Lourentzou 7/8/2026

ELSA3D: Elastic Semantic Anchoring for Unified 3D Understanding and Generation

ELSA3D unified 3D foundation model with elastic semantic anchoring for 3D generation and understanding.

Ax Zhengdao Chen, Eric Vanden-Eijnden, Joan Bruna 7/8/2026

A Functional-Space Mean-Field Theory of Partially-Trained Three-Layer Neural Networks

Mean-field theory analysis of three-layer neural network training dynamics and convergence properties.

Ax Abanoub Ghobrial, Kerstin Eder 7/8/2026

DIRA-SS:Dynamic Domain Incremental Regularised Adaptation -- Self-Supervised

DIRA-SS self-supervised adaptation for DNNs under domain shift in autonomous systems without retraining.

Ax Yawen Li, Yan Li, Junping Du, Yingxia Shao, Meiyu Liang, Guanhua Ye 7/8/2026

Trust-free Personalized Decentralized Learning

Trust-free personalized decentralized learning framework for federated settings without centralized coordinators.