Ax Boao Kong, Junzhu Liang, Yuxi Liu, Renjia Deng, Kun Yuan 5/14/2026

CR-Net: Scaling Parameter-Efficient Training with Cross-Layer Low-Rank Structure

CR-Net enables parameter-efficient LLM pre-training using cross-layer low-rank structure with reduced memory and computation.

Ax Zongyue Li, Xiao Han, Yusong Li, Niklas Strauss, Matthias Schubert 5/14/2026

DAWM: Diffusion Action World Models for Offline Reinforcement Learning via Action-Inferred Transitions

Diffusion-based world model for offline reinforcement learning that jointly generates actions, states, and rewards.

Ax Abhishek Moturu, Muhammad Muzammil, Anna Goldenberg, Babak Taati 5/14/2026

LiLAW: Lightweight Learnable Adaptive Weighting to Learn Sample Difficulty & Improve Noisy Training

LiLAW method that dynamically adjusts training sample weights based on difficulty for noisy data training.

Ax Zheng Wang, Kaixuan Zhang, Wanfang Chen, Xiaonan Lu, Longyuan Li, Tobias Schlagenhauf 5/14/2026

Characteristic Root Analysis and Regularization for Linear Time Series Forecasting

Theoretical analysis of linear models for time series forecasting, examining robustness and interpretability.

Ax Kamel Alrashedy, Vriksha Srihari, Zulfiqar Zaidi, Ridam Srivastava, Pradyumna Tambwekar, Matthew Gombolay 5/14/2026

Constraints-of-Thought: A Framework for Constrained Reasoning in Language-Model-Guided Search

Framework enabling LLMs to perform multi-step constrained reasoning by satisfying symbolic constraints in planning tasks.

Ax Mike Lasby, Ivan Lazarevich, Nish Sinnadurai, Sean Lie, Yani Ioannou, Vithursan Thangarasa 5/14/2026

REAP the Experts: Why Pruning Prevails for One-Shot MoE compression

Research on expert pruning vs merging for compressing Mixture-of-Experts models, showing pruning superior for generative tasks.

Ax Dario Shariatian, Alain Durmus, Umut Simsekli, Stefano Peluchetti 5/14/2026

Latent-Augmented Discrete Diffusion Models

Latent-Augmented Discrete Diffusion: learnable auxiliary channels for improved few-step language generation.

Ax Song Bian, Tao Yu, Shivaram Venkataraman, Youngsuk Park 5/14/2026

Scaling Laws Meet Model Architecture: Toward Inference-Efficient LLMs

Study of architectural factors and scaling laws trade-offs for inference-efficient LLMs.

Ax Mingyu Jo, Jaesik Yoon, Justin Deschenaux, Caglar Gulcehre, Sungjin Ahn 5/14/2026

Loopholing Discrete Diffusion: Deterministic Bypass of the Sampling Wall

Loopholing: mechanism preserving distributional information in discrete diffusion for parallel decoding.

Ax Yi He, Xingyu Zhou 5/14/2026

On the Sample Complexity of Differentially Private Policy Optimization

Theoretical analysis of sample complexity in differentially private policy optimization for RL.

Ax Tue Le, Linh Ngo Van, Trung Le 5/14/2026

Sharpness-Guided Group Relative Policy Optimization via Probability Shaping

Sharpness-guided group relative policy optimization for improving LLM reasoning via reinforcement learning.

Ax Marios Papamichalis, Regina Ruane 5/14/2026

Radial Compensation: Fixing Radius Distortion in Chart-Based Generative Models on Riemannian Manifolds

Radial compensation technique addressing radius distortion in chart-based generative models on Riemannian manifolds.

Ax Linnea M. Wolniewicz, Halil S. Kelebek, Simone Mestici, Michael D. Vergalla, Giacomo Acciarini, Bala Poduval, Olga Verkhoglyadova, Madhulika Guhathakurta, Thomas E. Berger, At{\i}l{\i}m G\"une\c{s} Baydin, Frank Soboczenski 5/14/2026

Connecting the Dots: A Machine Learning Ready Dataset for Ionospheric Forecasting Models

Open-access ionospheric dataset integrating diverse space weather observations for ML-based forecasting.

Ax Aniketh Iyengar, Jiaqi Han, Boris Ruf, Vincent Grari, Marcin Detyniecki, Stefano Ermon 5/14/2026

Energy Scaling Laws for Diffusion Models: Quantifying Compute in Image Generation

Energy scaling laws for diffusion models predicting computational demands across configurations.

Ax Ziqing Wen, Jiahuan Wang, Ping Luo, Dongsheng Li, Tao Sun 5/14/2026

FOAM: Blocked State Folding for Memory-Efficient LLM Training

FOAM: memory-efficient LLM training via blocked state folding to reduce Adam optimizer overhead.

Ax George Yakushev, Nataliia Babina, Masoud Vahid Dastgerdi, Vyacheslav Zhdanovskiy, Denis Kuznedelev, Alina Shutova, Max Ryabinin 5/14/2026

Asynchronous Reasoning: Training-Free Interactive Thinking LLMs

Training-free asynchronous reasoning for LLM agents enabling real-time responses without sequential thinking bottlenecks.

Ax Patrick Batsell, Satoshi Tsutsui, Bihan Wen 5/14/2026

Benchmarking Attribute Discrimination in Infant-Scale Vision-Language Models

Benchmark for evaluating attribute discrimination in infant-scale vision-language models on color, size, and texture.

Ax Zihan Yao, Ruoyu Wu, Tianxiang Gao 5/14/2026

Feature Learning Dynamics in Infinite-Depth Neural Networks

Mechanistic study of feature evolution during training in infinite-depth ResNets under depth-μP scaling.

Ax Yifan Zhang, Yifeng Liu, Mengdi Wang, Quanquan Gu 5/14/2026

Deep Delta Learning

Deep Delta Learning: transformer residual update rule enabling selective rewriting of content while preserving identity paths.

Ax Felix Biggs, Samuel Willis 5/14/2026

LLM Flow Processes for Text-Conditioned Regression

Flow-based processes for LLM regression tasks addressing error cascades and computational intensity in time-series and text-conditioned predictions.

Ax Antonio \'Alvarez-L\'opez, Borjan Geshkovski, Dom\`enec Ruiz-Balet 5/14/2026

Perceptrons and localization of attention's mean-field landscape

Theoretical study of transformer forward passes as interacting particle systems with gradient flow interpretation and mean-field limits.

Ax Mutian Yang, Zisen Zhan, Yutong Chen, Haolin Li, Kaiwen Wang, Kaili Zheng, Yuguang Wang, Qi Wang, Jiandong Gao, Ji Wu 5/14/2026

Collaborative Parameter Learning: Mitigating Forgetting via Parameter-Level Gradient Analysis

Proposes parameter-level gradient analysis to mitigate catastrophic forgetting during knowledge injection in large language models.

Ax Kaustubh Sharma, Srijan Tiwari, Ojasva Nema, Parikshit Pareek 5/14/2026

Mechanistic Evidence for Spectral Structures in Prior-Data Fitted Networks

Mechanistic study showing Prior-Data Fitted Networks learn structured spectral representations extractable as explicit kernels for amortized Bayesian inference.

Ax Puyu Wang, Junyu Zhou, Philipp Liznerski, Marius Kloft 5/14/2026

Optimization, Generalization and Differential Privacy Bounds for Gradient Descent on Kolmogorov-Arnold Networks

Theoretical analysis of gradient descent on Kolmogorov-Arnold Networks with bounds on optimization, generalization, and differential privacy properties.

Ax Markus Mueller, Kathrin Gruber, Dennis Fok 5/14/2026

Cascaded Flow Matching for Heterogeneous Tabular Data with Mixed-Type Features

Cascaded flow matching approach for generative modeling of heterogeneous tabular data with mixed discrete and continuous features.

Ax Idan Barnea, Orin Levy, Yishay Mansour 5/14/2026

The Horizon Threshold in Cooperative Multi-Agent Reward-Free Exploration

Theoretical analysis of cooperative multi-agent reward-free exploration in tabular finite-horizon MDPs with phased learning framework.

Ax Renukanandan Tumu, Aditya Singh, Rahul Mangharam 5/14/2026

AdaptNC: Adaptive Nonconformity Scores for Conformal Prediction under Distribution Shift

Adaptive conformal prediction method for uncertainty quantification under distribution shift in autonomous systems without exchangeability assumptions.

Ax Yoonjun Cho, Dongjae Jeon, Soeun Kim, Moongyu Jeon, Albert No 5/14/2026

Preserve-Then-Quantize: Balancing Rank Budgets for Quantization Error Reconstruction in LLMs

Proposes Preserve-Then-Quantize method for post-training quantization in LLMs, balancing rank budgets between preserving intrinsic structure and error reconstruction.

Ax Ibrahim Albool, Malak Gamal El-Din, Salma Elmalaki, Yasser Shoukry 5/14/2026

Exponential Approximation Rates and Parameter Efficiency of Learnable Bernstein Activations

Theoretical analysis of DeepBern-Nets using learnable Bernstein polynomial activations with guarantees on approximation error and parameter efficiency.

Ax Atula Tejaswi, Litu Rout, Constantine Caramanis, Sanjay Shakkottai, Sujay Sanghavi 5/14/2026

Entropy Aware Reward Guidance for Diffusion Language Model Alignment

Introduces EntRGi for reward guidance in discrete diffusion language models, enabling test-time adaptation without differentiating through discrete tokens.

Ax Yingke Li, Anjali Parashar, Enlu Zhou, Chuchu Fan 5/14/2026

Pragmatic Curiosity: A Unified Framework for Hybrid Learning and Optimization via Active Inference

Unified framework combining Bayesian optimization and experimental design via active inference for hybrid learning and optimization workflows.

Ax Fairoz Nower Khan, Nabuat Zaman Nahim, Ruiquan Huang, Haibo Yang, Peizhong Ju 5/14/2026

Flow Matching for Offline Reinforcement Learning with Discrete Actions

Extends flow matching to offline reinforcement learning with discrete actions and multiple objectives using continuous-time Markov chains.

Ax Adam Barla, Emanuele Nevali, Luca Viano, Volkan Cevher 5/14/2026

Provably avoiding over-optimization in Direct Preference Optimization without knowing the data distribution

Proposes PEPO, single-step direct preference optimization algorithm using pessimistic ensemble to avoid over-optimization in LLM preference learning without explicit reward models.

Ax Jingyao Wang, Peizheng Guo, Wenwen Qiang, Jiahuan Zhou, Huijie Guo, Changwen Zheng, Hui Xiong 5/14/2026

Towards Generalizable Reasoning: Group Causal Counterfactual Policy Optimization for LLM Reasoning

Introduces group causal counterfactual policy optimization to improve LLM reasoning generalization by crediting sound reasoning processes, not just correct answers.

Ax Manh Cuong Dao, Quang Hung Pham, Phi Le Nguyen, Thao Nguyen Truong, Bryan Kian Hsiang Low, Trong Nghia Hoang 5/14/2026

Diffusion-Inspired Reconfiguration of Transformers for Uncertainty Calibration

Proposes diffusion-inspired probabilistic transformer architecture for uncertainty calibration in risk-sensitive applications.

Ax Tatsuya Sagawa, Ryosuke Kojima 5/14/2026

How Well Do Large-Scale Chemical Language Models Transfer to Downstream Tasks?

Systematic evaluation of chemical language model scaling across model size, dataset size, and compute on molecular property prediction downstream tasks.

Ax Chendi Qian, Christopher Morris, Stefanie Jegelka, Christian Sohler 5/14/2026

Learning to Approximate Uniform Facility Location via Graph Neural Networks

Uses message-passing neural networks as heuristics for combinatorial optimization problems, combining learning with approximation algorithm guarantees.

Ax Nicolas Zumarraga, Thomas Kaar, Ning Wang, William Tennien, Alpay Hasanli, Max Rosenblattl, Fan Wu, Kevin Riehl, Maxwell A. Xu, Markus Kreft, Kevin O'Sullivan, Elgar Fleisch, Paul Schmiedmayer, Robert Jakob, Patrick Langer 5/14/2026

TS-Haystack: A Multi-Task Retrieval Benchmark for Long-Context Time-Series Reasoning

TS-Haystack: benchmark for evaluating time-series language models on retrieval and reasoning over long temporal contexts across 10 tasks.

Ax Junchen Liu, Sven Elflein, Or Litany, Zan Gojcic, Ruilong Li 5/14/2026

Test-Time Training with KV Binding Is Secretly Linear Attention

Research showing test-time training with KV binding is mathematically equivalent to learned linear attention, contradicting memorization interpretation.

Ax Alex Morehead, Miruna Cretu, Antonia Panescu, Rishabh Anand, Maurice Weiler, Tynan Perez, Samuel Blau, Steven Farrell, Wahid Bhimji, Anubhav Jain, Hrushikesh Sahasrabuddhe, Pietro Lio, Tommi Jaakkola, Rafael Gomez-Bombarelli, Rex Ying, N. Benjamin Erichson, Michael W. Mahoney 5/14/2026

Zatom-1: Towards a Multimodal Foundation Model for 3D Molecules and Materials

Zatom-1: multimodal foundation model unifying generative and predictive learning for 3D molecules and materials across domains.

Ax Cenwei Zhang, Lin Zhu, Manxi Lin, Lei You 5/14/2026

From Baselines to Transport Geodesics: Axiomatic Attribution via Optimal Generative Flows

Research on feature attribution via optimal transport flows, studying path ambiguity in counterfactual explanations.

Ax Ivan Luiz De Moura Matos, Abdel Djalil Sad Saoud, Ekaterina Iakovleva, Vito Paolo Pastore, Enzo Tartaglione 5/14/2026

Bias In, Bias Out? Finding Unbiased Subnetworks in Vanilla Models

Research extracting fair bias-agnostic subnetworks from vanilla trained models without additional unbiased data.

Ax Suorong Yang, Fangjian Su, Hai Gan, Ziqi Ye, Jie Li, Baile Xu, Furao Shen, Soujanya Poria 5/14/2026

Data Agent: Learning to Select Data via End-to-End Dynamic Optimization

Data Agent learns dynamic data selection via end-to-end optimization to prioritize informative samples and accelerate training.

Ax Savannah L. Ferretti, Jerry Lin, Sara Shamekh, Jane W. Baldwin, Michael S. Pritchard, Tom Beucler 5/14/2026

Data-Driven Integration Kernels for Interpretable Nonlocal Operator Learning

Data-driven integration kernels framework for interpretable nonlocal operator learning in climate and weather modeling applications.

Ax Tomas Ruiz, Zhen Qin, Yifan Zhang, Xuyang Shen, Yiran Zhong, Mengdi Wang 5/14/2026

FlashSampling: Fast and Memory-Efficient Exact Sampling

FlashSampling fuses categorical sampling into LM-head matmul for fast exact sampling without materializing logits in large-vocabulary decoding.

Ax Masoumeh Shafieinejad, Xi He, Mahshid Alinoori, John Jewell, Sana Ayromlou, Wei Pang, Veronica Chatrath, Gauri Sharma, Deval Pandya 5/14/2026

MIDST Challenge at SaTML 2025: Membership Inference over Diffusion-models-based Synthetic Tabular data

Challenge on membership inference attacks against synthetic tabular data generated by diffusion models for privacy evaluation.

Ax Ian Osband 5/14/2026

Delightful Distributed Policy Gradient

Research on distributed reinforcement learning addressing negative learning from high-surprisal data in stale or mismatched actor settings.

Ax Shenyang Deng, Zhuoli Ouyang, Tianyu Pang, Zihang Liu, Ruochen Jin, Shuhua Yu, Yaoqing Yang 5/14/2026

RMNP: Row-Momentum Normalized Preconditioning for Scalable Matrix-Based Optimization

Row-momentum normalized preconditioning optimization method for scalable neural network training capturing curvature information efficiently.

Ax Zakaria Mhammedi, James Cohan 5/14/2026

Decoupling Exploration and Policy Optimization: Uncertainty Guided Tree Search for Hard Exploration

Research on uncertainty-guided tree search decoupling exploration from policy optimization for hard exploration RL problems.

Ax Xiang Li, Yixuan Jia, Xiao Li, Jeffrey A. Fessler, Rongrong Wang, Qing Qu 5/14/2026

MCLR: Improving Conditional Modeling via Inter-Class Likelihood-Ratio Maximization and Unifying Classifier-Free Guidance with Alignment Objectives

Research unifying classifier-free guidance with alignment objectives in diffusion models via inter-class likelihood-ratio maximization.