Ax Anders Gj{\o}lbye, Lars Kai Hansen, Sanmi Koyejo 5/18/2026

Reasoning Models Don't Just Think Longer, They Move Differently

Study of hidden-state trajectories in reasoning-trained LLMs showing they follow different internal paths during chain-of-thought.

Ax Libo Sun, Po-wei Harn, Peixiong He, Xiao Qin 5/18/2026

When Does Sparse MoE Help in Vision? The Role of Backbone Compute Leverage in Sparse Routing

Analysis of when sparse Mixture-of-Experts routing benefits vision models, identifying compute-leverage patterns.

Ax Guruprasad Raghavan, George Nychis, Rohan Narayana Murthy 5/18/2026

X-SYNTH: Beyond Retrieval -- Enterprise Context Synthesis from Observed Human Attention

X-SYNTH system for AI agents to synthesize enterprise context from human attention patterns rather than retrieval.

Ax Yufeng Du, Phillip Harris, Minyang Tian, Eliu A Huerta, Srikanth Ronanki, Subendhu Rongali, Aram Galstyan, Hao Peng 5/18/2026

RoPE Distinguishes Neither Positions Nor Tokens in Long Contexts, Provably

Theoretical analysis proving RoPE positional embeddings lose effectiveness in long-context LLMs.

Ax Jinyuan Li, Langlin Huang, Chengsong Huang, Shaoyang Xu, Donghong Cai, Yuyi Yang, Wenxuan Zhang, Jiaxin Huang 5/18/2026

Process Rewards with Learned Reliability

Distributional process reward model predicting step-level success probability and reliability for reasoning tasks.

Ax De Shuai Zhang 5/18/2026

When Latent Geometry Is Not Enough: Draft-Conditioned Latent Refinement for Non-Autoregressive Text Generation

Non-autoregressive text generation model using draft-conditioned latent refinement with flow networks.

Ax Nurbek Tastan, Alex Iacob, Lorenzo Sani, Meghdad Kurmanji, Nicholas D. Lane, Samuel Horvath, Karthik Nandakumar 5/18/2026

Response-Conditioned Parallel-to-Sequential Orchestration for Multi-Agent Systems

Multi-agent orchestration framework dynamically switching between parallel and sequential LLM agent collaboration.

Ax Fengfei Yu, Ruijia Niu, Dongxia Wu, Yian Ma, Rose Yu 5/18/2026

Calibrating LLMs with Semantic-level Reward

Calibration method for LLMs using semantic-level rewards to improve uncertainty estimation in high-stakes tasks.

Ax Vinayshekhar Bannihatti Kumar, Disha Makhija, Manoj Ghuhan Arivazhagan, Rashmi Gangadharaiah 5/18/2026

Syntax Without Semantics: Teaching Large Language Models to Code in an Unseen Language

Study evaluating LLM code generation transfer to unseen programming languages through fine-tuning analysis.

Ax Senmao Tian, Xiang Wei, Shunli Zhang 5/18/2026

Neutral-Reference Prompting for Vision-Language Models

Prompting technique for vision-language models addressing base-new class trade-off in transfer learning.

Ax Yilong Wan, Yuqiang Li, Xianyi Wu 5/18/2026

Pessimistic Risk-Aware Policy Learning in Contextual Bandits

Offline policy learning framework for contextual bandits optimizing under general risk criteria.

Ax Liqi Zhou, Jiafu Li 5/18/2026

Few-Shot Large Language Models for Actionable Triage Categorization of Online Patient Inquiries

Few-shot LLM approach for triaging online patient inquiries into clinical action categories with minimal labeled data.

Ax Ekkehard Schnoor, Jawher Said, Malik Tiomoko, Wojciech Samek, Alexander Jung 5/18/2026

$\alpha$-TCAV: A Unified Framework for Testing with Concept Activation Vectors

Statistical framework improving Concept Activation Vectors for interpretability in deep learning models.

Ax Himanshu Singh Baghel 5/18/2026

ADAPT: A Self-Calibrating Proactive Autoscaler for Container Orchestration

Adaptive autoscaler for container orchestration that learns cold-start duration using EWMA estimation.

Ax Sujie Hu, Chubin Chen, Jiashu Zhu, Jiahong Wu, Xiangxiang Chu, Xiu Li 5/18/2026

Embedding-perturbed Exploration Preference Optimization for Flow Models

Preference optimization framework for flow models addressing intra-group variance decay in RL alignment of generative models.

Ax J\'ulia Orteu Aubach, Fabio Banchelli, Marc Clasc\`a Ram\'irez, Marta Garcia-Gasulla 5/18/2026

Heuristic-Based Merging of HPC Traces to Extend Hardware Counter Coverage

ML framework for HPC performance prediction using merged execution traces to overcome hardware counter limitations.

Ax Syed Waqas Ali, Ibrar Ali Shah, Farzana Zahid, Daniyal Munir, Hans D. Schotten 5/18/2026

A Multi-Layer Cloud-IDS Pipeline with LLM and Adaptive Q-Learning Calibration

Multi-layer cloud intrusion detection system combining LLMs with Q-learning for improved performance in real deployments.

Ax Sebastian Springer, Alessandro Laio 5/18/2026

Unsupervised Domain Shift Detection with Interpretable Subspace Attribution

Tool for detecting and interpreting domain shifts in high-dimensional data by identifying anomalies in feature subspaces.

Ax Nils Feldhus, Tanja Baeumel, Elena Golimblevskaia, Qianli Wang, Van Bach Nguyen, Aaron Louis Eidt, Christopher Ebert, Wojciech Samek, Jing Yang, Vera Schmitt, Sebastian M\"oller, Simon Ostermann 5/18/2026

Judge Circuits

Causal analysis of format inconsistencies in LLM-as-judge scoring using PEAP to investigate internal mechanisms.

Ax Zijie Dai, Shiyuan Deng, Sheng Guan, Yizhou Tian, Xin Yao, Xiao Yan, James Cheng 5/18/2026

RecMem: Recurrence-based Memory Consolidation for Efficient and Effective Long-Running LLM Agents

RecMem: memory consolidation system for long-running LLM agents that reduces token consumption through lazy consolidation.

Ax Augusto B. Corr\^ea, Andr\'e G. Pereira, Jendrik Seipp 5/18/2026

Property-Guided LLM Program Synthesis for Planning

Uses property-guided synthesis to reduce LLM inference costs in program synthesis for planning, guiding generation with formal properties.

Ax Yishun Lu, Junhao Zhang, Zeyu Yang, Wes Armour 5/18/2026

Runtime-Orchestrated Second-Order Optimization for Scalable LLM Training

Asteria: runtime system for scalable LLM training using second-order optimization, decoupling preconditioner state from GPU path.

Ax Parand A. Alamdari, Toryn Q. Klassen, Sheila A. McIlraith 5/18/2026

Formal Methods Meet LLMs: Auditing, Monitoring, and Intervention for Compliance of Advanced AI Systems

Combines formal methods with ML for auditing and monitoring AI systems across development lifecycle, enabling compliance checks.

Ax Igor Bogdanov, Chung-Horng Lung, Thomas Kunz, Jie Gao, Adrian Taylor, Marzia Zaman 5/18/2026

Context, Reasoning, and Hierarchy: A Cost-Performance Study of Compound LLM Agent Design in an Adversarial POMDP

Controlled study of compound LLM agent design trade-offs in adversarial environments, analyzing context, reasoning, and task decomposition.

Ax Nathan Roll, Jill Kries, Laura Gwilliams, Cory Shain 5/18/2026

Artificial Aphasias in Lesioned Language Models

Technique to characterize language model organization by lesioning parameters, inspired by neuroscience aphasia studies.

Ax Pavan Manjunath, Thomas pruefer 5/18/2026

A Unified Generative-AI Framework for Smart Energy Infrastructure: Intelligent Gas Distribution, Utility Billing, Carbon Analytics, and Quantum-Inspired Optimisation

arXiv: Framework combining generative AI, smart metering, quantum optimization for energy infrastructure and billing.

Ax Igor Bogdanov, Chung-Horng Lung, Thomas Kunz, Jie Gao, Adrian Taylor, Marzia Zaman 5/18/2026

FORGE: Self-Evolving Agent Memory With No Weight Updates via Population Broadcast

arXiv: FORGE—population-based protocol for self-evolving LLM agent memory without gradient updates. ReAct agents with Reflexion.

Ax Stratis Tsirtsis, Kai Rawal, Chris Russell, Brent Mittelstadt, Sandra Wachter 5/18/2026

AI-Mediated Communication Can Steer Collective Opinion

arXiv: Studies how LLM-mediated communication influences collective opinion formation on social platforms.

Ax Yongcun Song, Ziqi Wang, Enrique Zuazua 5/18/2026

Approximate and Weighted Data Reconstruction Attack in Federated Learning

arXiv: Data reconstruction attacks against federated learning systems. Privacy and security analysis of FL protocols.

Ax Leo Muxing Wang, Pengkun Yang, Lili Su 5/18/2026

On the Convergence Rates of Federated Q-Learning across Heterogeneous Environments

arXiv: Convergence analysis of federated Q-learning in heterogeneous multi-agent environments. Theory for distributed RL.

Ax Leonardo Ferreira Guilhoto, Paris Perdikaris 5/18/2026

Deep Learning Alternatives of the Kolmogorov Superposition Theorem

arXiv: Explores Kolmogorov Superposition Theorem alternatives for neural network design beyond Kolmogorov-Arnold Networks.

Ax Pritam Anand, Tathagata Bandyopadhyay, Suresh Chandra 5/18/2026

Tube Loss: A Novel Approach for Prediction Interval Estimation

arXiv: Tube Loss function for prediction interval estimation in regression. Novel loss function with theoretical guarantees.

Ax Wenyun Li, Wenjie Huang, Chen Sun 5/18/2026

Shaping Sparse Rewards in Reinforcement Learning: A Semi-supervised Approach

arXiv: Semi-supervised learning approach for sparse reward shaping in reinforcement learning. Uses SSL and data augmentation.

Ax Ruicheng Ao, Gan Luo, David Simchi-Levi, Xinshang Wang 5/18/2026

Optimizing LLM Inference: Fluid-Guided Online Scheduling with Memory Constraints

GPU scheduling algorithm for LLM inference that manages KV cache memory constraints and minimizes latency under $700k daily inference costs.

Ax Yang Li 5/18/2026

Rethinking Predictive Modeling for LLM Routing: When Simple kNN Beats Complex Learned Routers

Comparison of LLM routing strategies showing simple k-NN outperforms complex learned routers for selecting specialized models.

Ax Yuantu Zhu, Zheyan Li, Dai Shi, Luke Thompson, Oliver Nash, Jose Miguel Lara Rangel, Siran Li, Bingguang Chen, Rongchan Zhu, Qi Meng, Hao Ni 5/18/2026

SPDEBench: An Extensive Benchmark for Learning Stochastic PDEs

Comprehensive benchmark for learning surrogate models of stochastic PDEs with complex spatio-temporal dynamics.

Ax Xiaoqiang Lin, Arun Verma, Zhongxiang Dai, Daniela Rus, See-Kiong Ng, Bryan Kian Hsiang Low 5/18/2026

ActiveDPO: Active Direct Preference Optimization for Sample-Efficient Alignment

Active learning approach for LLM alignment that selectively samples preference annotations, reducing cost while maintaining alignment quality.

Ax Andreas Bergmeister, Manish Krishan Lal, Stefanie Jegelka, Suvrit Sra 5/18/2026

A projection-based framework for gradient-free and parallel learning

Proposes gradient-free neural network training via projection operators and feasibility-seeking, alternative to conventional loss minimization.

Ax Prabhat Nagarajan, Martha White, Marlos C. Machado 5/18/2026

Deep Double Q-learning

Extends double Q-learning to deep RL, improving target bootstrap decoupling in value function estimation.

Ax Zeyu Huang, Tianhao Cheng, Zihan Qiu, Zili Wang, Yinghui Xu, Edoardo M. Ponti, Ivan Titov 5/18/2026

Blending Supervised and Reinforcement Fine-Tuning with Prefix Sampling

Combines supervised and reinforcement fine-tuning for LLMs using prefix sampling, addressing trade-offs between behavior cloning and performance gains.

Ax Mehul Damani, Isha Puri, Stewart Slocum, Idan Shenfeld, Leshem Choshen, Yoon Kim, Jacob Andreas 5/18/2026

Beyond Binary Rewards: Training LMs to Reason About Their Uncertainty

Trains LMs via RL to reason about uncertainty rather than just correctness, improving performance on question answering by penalizing low-confidence outputs.

Ax Xing Wei, Yuanhang Wang, Duoxiang Zhao, Zezhou Zhang, Hao Qin, Yuqi Ouyang 5/18/2026

Generalized Policy Gradient with History-Aware Decision Transformer for Reliable Routing over Graph Signals

Applies policy gradient with decision transformers for reliable path planning in stochastic transportation networks with uncertain travel times.

Ax George Yakushev, Alina Shutova, Ivan Rubachev, Natalia Bereberdina, Renat Sergazinov, Artem Babenko 5/18/2026

Talking Trees: Reasoning-Assisted Induction of Decision Trees for Tabular Data

Uses LLM reasoning to improve decision tree induction for tabular data, balancing interpretability with performance over black-box foundation models.

Ax Jo\~ao Paulo Vieira, Victor Afonso Bauler, Rodrigo Kobashikawa Rosa, Danilo Silva 5/18/2026

Towards a more realistic evaluation of machine learning models for bearing fault diagnosis

Examines data leakage issues in ML models for bearing fault diagnosis, highlighting methodological flaws preventing real-world generalization in condition monitoring.

Ax Parth Asawa, Alan Zhu, Abigail O'Neill, Matei Zaharia, Alexandros G. Dimakis, Joseph E. Gonzalez 5/18/2026

How to Train Your Advisor: Steering Black-Box LLMs with Advisor Models

Method to train small open-weight advisor models that generate dynamic prompts to improve black-box LLM performance. Demonstrates 27.4% improvement on GPT-5.2 tax tasks.

Ax F\'elix Vandervorst, Bruno Deprez, Wouter Verbeke, Tim Verdonck 5/18/2026

Inductive inference of gradient-boosted decision trees on graphs for insurance fraud detection

Graph-based gradient boosting approach for insurance fraud detection using inductive inference on networked claim data.

Ax Ruizhe Wang, Yucheng Ding, Xiao Liu, Yaoxiang Wang, Peng Cheng, Baining Guo, Zhengjun Zha, Yeyun Gong 5/18/2026

Beyond Sunk Costs: Boosting LLM Pre-training Efficiency via Orthogonal Growth of Mixture-of-Experts

Strategy for efficiently pre-training large LLMs by orthogonally expanding Mixture-of-Experts parameters from existing checkpoints.

Ax Mengqi Li, Lei Zhao, Anthony Man-Cho So, Ruoyu Sun, Xiao Li 5/18/2026

A Model Can Help Itself: Reward-Free Self-Training for LLM Reasoning

Post-training method enabling LLMs to improve reasoning without external rewards through self-generation and self-training.

Ax Jamison Meindl, Yunsheng Tian, Tony Cui, Veronika Thost, Zhang-Wei Hong, Jie Chen, Wojciech Matusik, Mina Konakovi\'c Lukovi\'c 5/18/2026

SemanticOpt: Towards LLM-Based Semantic Black-Box Optimization

LLM-based semantic optimization approach for expensive black-box problems incorporating domain knowledge and heuristics.

Ax Vincent-Daniel Yun, Junhyuk Jo, Sunwoo Lee 5/18/2026

Weight Concentration Regularization for Improving Pruning Robustness Under High Sparsity

Regularization technique for neural network pruning that improves robustness under aggressive sparsity levels.