Ax Yuheng Luo, Shuyan Li, Zhong Cao 5/1/2026

GRASP: group-Shapley feature selection for patients

GRASP framework combining Shapley values and group L21 regularization for interpretable feature selection in medical prediction.

Ax Arjun Chatterjee, Sayeed Sajjad Razin, John Wu, Siddhartha Laghuvarapu, Jathurshan Pradeepkumar, Jimeng Sun 5/1/2026

Making Conformal Predictors Robust in Healthcare Settings: a Case Study on EEG Classification

Conformal prediction methods for robust uncertainty quantification in clinical EEG classification under distribution shift.

Ax Frederic Wang, Katherine L. Bouman 5/1/2026

Sample-efficient evidence estimation of score based priors for model selection

Sample-efficient model evidence estimation using score-based priors and diffusion models for Bayesian inverse imaging problems.

Ax Thomas R\"uckstie{\ss}, Robin Vujanic 5/1/2026

Autoregressive Synthesis of Sparse and Semi-Structured Mixed-Type Data

Autoregressive method for generating synthetic sparse and semi-structured JSON data with nested objects and variable-length arrays.

Ax Bojian Yin, Shurong Wang, Haoyu Tan, Sander Bohte, Federico Corradi, Guoqi Li 5/1/2026

Efficient Sparse Selective-Update RNNs for Long-Range Sequence Modeling

Sparse selective-update RNNs reducing memory decay for efficient long-range sequence modeling in audio/video.

Ax Zhengzhao Ma, Xueru Wen, Boxi Cao, Yaojie Lu, Hongyu Lin, Jinglin Yang, Min He, Xianpei Han, Le Sun 5/1/2026

Decoupling Reasoning and Confidence: Resurrecting Calibration in Reinforcement Learning from Verifiable Rewards

Decoupling reasoning and confidence in LLMs through gradient conflict analysis for improved calibration in verifiable reward learning.

Ax Selim An, Il hong Suh, Yeseong Kim 5/1/2026

GlowQ: Group-Shared LOw-Rank Approximation for Quantized LLMs

Low-rank correction method for quantized LLMs reducing accuracy degradation in 4-bit representations with minimal overhead.

Ax Qingxiu Liu, Cyril Y. He, Hanser Jiang, Zion Wang, Alan Zhao, Patrick P. C. Lee 5/1/2026

FluxMoE: Decoupling Expert Residency for High-Performance MoE Serving

Expert residency decoupling optimization for efficient MoE LLM serving with improved KV cache utilization.

Ax Zhaowen Fan, Rongchao Zhang 5/1/2026

Event-Centric World Modeling with Memory-Augmented Retrieval for Embodied Decision-Making

Memory-augmented event-centric world modeling framework for embodied AI agents in safety-critical environments.

Ax Meghana Kshirsagar, Allen Nie, Ching-An Cheng, Fanglei Xue, Rahul Dodhia, Juan Lavista Ferres, Kevin K. Yang, Frank DiMaio 5/1/2026

RosettaSearch: Multi-Objective Inference-Time Search for Protein Sequence Design

Multi-objective inference-time search using LLMs as optimizer for protein sequence design with structure prediction rewards.

Ax Yunke Ao, Le Chen, Bruce D. Lee, Assefa S. Wahd, Aline Czarnobai, Philipp F\"urnstahl, Bernhard Sch\"olkopf, Andreas Krause 5/1/2026

Bounded Ratio Reinforcement Learning

Bounded Ratio Reinforcement Learning framework bridging trust region theory and PPO's clipped objective.

Ax Seoungbin Bae, Dabeen Lee 5/1/2026

Logistic Bandits with $\tilde{O}(\sqrt{dT})$ Regret without Context Diversity Assumptions

Logistic bandit algorithm achieving optimal regret bounds without context diversity assumptions.

Ax Kevin McKee, Thomas Hazy, Yicong Zheng, Zacharie Bugaud, Thomas Miconi 5/1/2026

Cortex-Inspired Continual Learning: Unsupervised Instantiation and Recovery of Functional Task Networks

Parameter isolation method inspired by neocortex structure for continual learning without catastrophic forgetting.

Ax Jiachen Liu, Jiaxin Pei, Jintao Huang, Chenglei Si, Ao Qu, Xiangru Tang, Runyu Lu, Lichang Chen, Xiaoyan Bai, Haizhong Zheng, Carl Chen, Zhiyang Chen, Haojie Ye, Yujuan Fu, Zexue He, Zijian Jin, Zhenyu Zhang, Shangquan Sun, Maestro Harmon, John Dianzhuo Wang, Jianqiao Zeng, Jiachen Sun, Mingyuan Wu, Baoyu Zhou, Chenyu You, Shijian Lu, Yiming Qiu, Fan Lai, Yuan Yuan, Yao Li, Junyuan Hong, Ruihao Zhu, Beidi Chen, Alex Pentland, Ang Chen, Mosharaf Chowdhury, Zechen Zhang 5/1/2026

The Last Human-Written Paper: Agent-Native Research Artifacts

Research on AI agents as native artifacts that replace traditional linear scientific papers with branching exploration records.

Ax Martin G. Frasch 5/1/2026

minAction.net: Energy-First Neural Architecture Design -- From Biological Principles to Systematic Validation

Energy-aware neural architecture design evaluated across 2,203 experiments incorporating biological constraints into ML optimization.

Ax Ching-Chun Chang, Kai Gao, Shuying Xu, Anastasia Kordoni, Christopher Leckie, Isao Echizen 5/1/2026

Hypnopaedia-Aware Machine Unlearning via Psychometrics of Artificial Mental Imagery

Machine unlearning approach to remove neural backdoors and defend against trigger-based attacks in learning systems.

Ax Sakshi Agarwal, Gabriel Hope, Jimin Heo, Erik B. Sudderth 5/1/2026

VIPaint: Image Inpainting with Pre-Trained Diffusion Models via Variational Inference

Variational inference approach for conditioning diffusion models on masked images to improve inpainting sample quality.

Ax Daniel Andrew Coulson, Martin T. Wells 5/1/2026

Foreclassing: A new machine learning perspective on human decision making with temporal data

Foreclassing framework automates human decision-making by integrating time series forecasts with prior experience and outcome uncertainty.

Ax Musfiqur Rahman, SayedHassan Khatoonabadi, Emad Shihab 5/1/2026

OpenClassGen: A Large-Scale Corpus of Real-World Python Classes for LLM Research

OpenClassGen dataset of 324,843 Python classes from 2,970 open-source projects for evaluating and training LLM-based code generation models.

Ax Cyril Shih-Huan Hsu, Anestis Dalgkitsis, Chrysa Papagianni, Paola Grosso 5/1/2026

Transformer-Empowered Actor-Critic Reinforcement Learning for Sequence-Aware Service Function Chain Partitioning

Actor-critic reinforcement learning with transformers for service function chain partitioning in 6G network virtualized network function management.

Ax Nan Lu, Ethan Lee, Ethan X. Fang, Junwei Lu 5/1/2026

Contextual Online Uncertainty-Aware Preference Learning for Human Feedback

Statistical framework for online uncertainty-aware preference learning from human feedback to align large models using contextual dynamic information.

Ax Zhuohan Xie, Daniil Orel, Rushil Thareja, Dhruv Sahnan, Hachem Madmoun, Fan Zhang, Debopriyo Banerjee, Georgi Georgiev, Xueqing Peng, Lingfei Qian, Jimin Huang, Jinyan Su, Aaryamonvikram Singh, Rui Xing, Rania Elbadry, Chen Xu, Haonan Li, Fajri Koto, Ivan Koychev, Tanmoy Chakraborty, Yuxia Wang, Salem Lahlou, Veselin Stoyanov, Sophia Ananiadou, Preslav Nakov 5/1/2026

FinChain: A Symbolic Benchmark for Verifiable Chain-of-Thought Financial Reasoning

FinChain benchmark evaluates chain-of-thought reasoning in financial analysis, emphasizing verifiable intermediate steps beyond final numerical answers.

Ax Hongyu H\`e, Minhao Jin, Maria Apostolaki 5/1/2026

Making Logic a First-Class Citizen in Generative ML for Networking

Integrates logic constraints into generative ML models for networking tasks to enforce rule compliance and improve trustworthiness and controllability.

Ax Santosh Vasa, Aditi Ramadwar, Jnana Rama Krishna Darabattula, Md Zafar Anwar, Stanislaw Antol, Andrei Vatavu, Thomas Monninger, Sihao Ding 5/1/2026

AutoVDC: Automated Vision Data Cleaning Using Vision-Language Models

AutoVDC framework uses vision-language models to automatically identify and clean annotation errors in autonomous driving datasets.

Ax Deniz Bayazit, Aaron Mueller, Antoine Bosselut 5/1/2026

Crosscoding Through Time: Tracking Emergence & Consolidation Of Linguistic Representations Throughout LLM Pretraining

Tracks emergence and consolidation of linguistic representations during LLM pretraining using sparse coding to understand concept acquisition timing.

Ax Savini Kashmira, Jayanaka Dantanarayana, Thamirawaran Sathiyalogeswaran, Krisztian Flautner, Lingjia Tang, Jason Mars 5/1/2026

GraphMend: Code Transformations for Fixing Graph Breaks in PyTorch 2

GraphMend compiler technique eliminates FX graph breaks in PyTorch 2, reducing fallbacks to eager mode and improving compilation performance.

Ax Saeed Ghadimi, Woosuk L. Jung, Arnesh Sujanani, David Torregrosa-Bel\'en, Henry Wolkowicz 5/1/2026

Optimal Diagonal Preconditioning Beyond Worst-Case Conditioning: Theory and Practice of Omega Scaling

Derives optimal diagonal preconditioning using omega-condition number with pseudoconvex reformulation for improved numerical optimization.

Ax Jiaqi Leng, Xiang Hu, Junxiong Wang, Jianguo Li, Wei Wu, Yucheng Lu 5/1/2026

Understanding and Improving Length Generalization in Hierarchical Sparse Attention Models

Analyzes length generalization in hierarchical sparse attention models for long-context processing, comparing against transformers and state space models.

Ax Ming Li, Pei Chen, Zhenhao Zhang, Tao Yang, Xinyang Zhang, Han Li, Tianyu Cao, Ming Zeng, Zhuofeng Wu, Meng Jiang, Huasheng Li, Lihong Li, Bing Yin 5/1/2026

Mitigating Lost in Multi-turn Conversation via Curriculum RL with Verifiable Accuracy and Abstention Rewards

RLAAR framework uses curriculum reinforcement learning with verifiable rewards to mitigate Lost-in-Conversation performance degradation in multi-turn LLM interactions.

Ax Parsa Rangriz 5/1/2026

Limit Theorems for Stochastic Gradient Descent in High-Dimensional Single-Layer Networks

Theoretical analysis of stochastic gradient descent scaling limits in high-dimensional single-layer networks, studying critical step size regimes.

Ax Meixiu Long, Duolin Sun, Dan Yang, Yihan Jiao, Lei Liu, Jiahai Wang, BinBin Hu, Yue Shen, Jie Feng, Zhehao Tan, Junjie Wang, Lianzhen Zhong, Jian Wang, Peng Wei, Jinjie Gu 5/1/2026

GroupRank: A Groupwise Paradigm for Effective and Efficient Passage Reranking with LLMs

GroupRank uses LLMs for passage reranking via groupwise paradigm, balancing efficiency and accuracy in information retrieval without pointwise limitations.

Ax Emma Andrews, Prabhat Mishra 5/1/2026

Quantum Masked Autoencoders for Vision Learning

Quantum masked autoencoders extend classical masked autoencoders to leverage quantum computing for improved feature learning in vision tasks.

Ax Leo Fillioux, Enzo Ferrante, Paul-Henry Courn\`ede, Maria Vakalopoulou, Stergios Christodoulidis 5/1/2026

PVeRA: Probabilistic Vector-Based Random Matrix Adaptation

PVeRA proposes probabilistic vector-based random matrix adaptation for efficient finetuning of large foundation models on small datasets and limited compute.

Ax Yonggan Fu, Lexington Whalen, Zhifan Ye, Xin Dong, Shizhe Diao, Jingyu Liu, Chengyue Wu, Hao Zhang, Enze Xie, Song Han, Maksim Khadkevich, Jan Kautz, Yingyan Celine Lin, Pavlo Molchanov 5/1/2026

Efficient-DLM: From Autoregressive to Diffusion Language Models, and Beyond in Speed

Method to convert pretrained autoregressive language models into efficient diffusion language models for faster parallel generation while maintaining task accuracy.

Ax Shir Rozenfeld, Rahul Pankajakshan, Itay Zloczower, Eyal Lenga, Gilad Gressel, Yisroel Mirsky 5/1/2026

GAVEL: Towards Rule-Based Safety Through Activation Monitoring

arXiv paper introducing rule-based activation monitoring for LLM safety detection with improved interpretability and flexibility.

Ax I-Chun Arthur Liu, Krzysztof Choromanski, Sandy Huang, Connor Schenck 5/1/2026

CLAMP: Contrastive Learning for 3D Multi-View Action-Conditioned Robotic Manipulation Pretraining

arXiv paper on contrastive learning for 3D multi-view action-conditioned robotic manipulation pretraining to capture spatial information.

Ax Lingwei Gu, Nour Jedidi, Jimmy Lin 5/1/2026

NanoKnow: How to Know What Your Language Model Knows

arXiv paper analyzing what language models know using NanoChat's open pre-training data to trace parametric knowledge sources.

Ax En-Ya Kuo, Sebastien Motsch 5/1/2026

EmDT: Embedding Diffusion Transformer for Tabular Data Generation in Fraud Detection

arXiv paper on diffusion transformer for generating synthetic fraudulent samples in imbalanced datasets using UMAP clustering.

Ax Jinquan Zheng, Jia Yuan, Jiacheng Yao, Chenyang Gu, Pujun Zheng, Guoxiu He 5/1/2026

Mitigating Selection Bias in Large Language Models via Permutation-Aware GRPO

arXiv paper addressing selection bias in LLMs for multiple-choice tasks via permutation-aware group relative policy optimization training.

Ax Tiantian Zhang, Jierui Zuo, Michael Chen, Wenping Wang 5/1/2026

DDO-RM: Distribution-Level Policy Improvement after Reward Learning

arXiv paper proposing DDO-RM, a reward learning method that converts reward scores to target distributions via KL-regularized mirror descent.

Ax Zhongkai Yu, Haotian Ye, Chenyang Zhou, Ohm Rishabh Venkatachalam, Zaifeng Pan, Zhengding Hu, Junsung Kim, Won Woo Ro, Po-An Tsai, Shuyi Pei, Yangwook Kang, Yufei Ding 5/1/2026

AMMA: A Multi-Chiplet Memory-Centric Architecture for Low-Latency 1M Context Attention Serving

arXiv paper proposing memory-centric architecture for serving 1M token LLM contexts with low latency by prioritizing memory over GPU compute.

HN tjek 5/1/2026

Stripe link-CLI: Let your agents spend on your behalf

Stripe Link CLI tool enabling AI agents to make secure payments with one-time-use credentials without storing card details.

HN zhoutong 5/1/2026

DeepSeek V4 Flash and V4 Pro in Microsoft Foundry

Microsoft Foundry adds DeepSeek V4 Flash and V4 Pro models for building cost-effective production AI systems.

HN nsoonhui 5/1/2026

AI Value Capture – The Shift to Model Labs

Analysis of AI value capture shift toward model labs, discussing agentic AI inflection point and token economics.

HN mindsaspire 5/1/2026

KV Cache Locality: The Hidden Variable in Your LLM Serving Cost

Technical analysis of KV cache locality inefficiencies in distributed LLM serving causing redundant computation and cost overhead.

HN softservo 5/1/2026

Text-to-CAD

Open source harness for generating 3D CAD models using coding agents like Claude and Codex with bundled skills.

HN gnabgib 5/1/2026

Enabling privacy-preserving AI training on everyday devices

MIT research on privacy-preserving AI training methods that run on everyday devices.

HN laxmena 5/1/2026

AI Skills as loader spec, not prompts – why the architecture changes everything

Architecture approach for AI agent skills as executable programs with loader specs instead of prompts, addressing software rot and reproducibility.

HN nclin_ 5/1/2026

Show HN: LLM-Powered News –> Event Map, Timeline, and Analysis

Open source LLM-powered pipeline extracting claims and events from news, synthesizing timelines and analysis with actor attribution.

LB kmjn.org by mjn 5/1/2026

Porting microgpt to Futhark, Part I

Porting microgpt (GPT-2 implementation) from Python to Futhark, a data-parallel language, to improve scalability and performance.