Ax Hengshuai Yao, Xing Chen, Ahmed Murtadha, Guan Wang 3/18/2026

Thin Keys, Full Values: Reducing KV Cache via Low-Dimensional Attention Selection

KV cache optimization using low-dimensional attention selection to reduce transformer memory with O(log N) key dimensions.

Ax Shubhangi Upasani, Chen Wu, Jay Rainton, Bo Li, Urmish Thakker, Changran Hu, Qizheng Zhang 3/18/2026

Test-Time Adaptation via Many-Shot Prompting: Benefits, Limits, and Pitfalls

Study examining many-shot prompting for test-time LLM adaptation, analyzing reliability and limits of in-context learning scaling.

Ax Yunni Qu (The University of North Carolina at Chapel Hill), Dzung Dinh (The University of North Carolina at Chapel Hill), Grant King (University of Michigan), Whitney Ringwald (University of Minnisota Twin Cities), Bing Cai Kok (The University of North Carolina at Chapel Hill), Kathleen Gates (The University of North Carolina at Chapel Hill), Aidan Wright (University of Michigan), Junier Oliva (The University of North Carolina at Chapel Hill) 3/18/2026

Relaxed Efficient Acquisition of Context and Temporal Features

Active feature acquisition method for biomedical applications optimizing measurement selection under temporal and cost constraints.

Ax Yuval Ran-Milo 3/18/2026

Attention Sinks Are Provably Necessary in Softmax Transformers: Evidence from Trigger-Conditional Tasks

Theoretical analysis proving attention sinks are functionally necessary in softmax transformers for certain tasks.

Ax Vincent Zhihao Zheng, \'Etienne Marcotte, Arjun Ashok, Andrew Robert Williams, Lijun Sun, Alexandre Drouin, Valentina Zantedeschi 3/18/2026

Overcoming the Modality Gap in Context-Aided Forecasting

Research addressing multimodal model underperformance in context-aided forecasting via improved context quality assessment.

Ax Yuzhang Xie, Yuhua Wu, Ruiyu Wang, Fadi Nahab, Xiao Hu, Carl Yang 3/18/2026

Enhanced Atrial Fibrillation Prediction in ESUS Patients with Hypergraph-based Pre-training

Machine learning method using hypergraph pre-training to improve atrial fibrillation prediction in stroke patients.

Ax Naaisha Agarwal, Yihan Wu, Yichang Jian, Yikuan Hu, Nishad Mansoor, Mohan Li, Yifei Peng, Wang-Zhou Dai, Yao-Xiang Ding, Emanuele Sansone 3/18/2026

OrigamiBench: An Interactive Environment to Synthesize Flat-Foldable Origamis

Interactive benchmark environment for synthesizing flat-foldable origamis, testing AI systems' planning and causal reasoning in physical domains.

Ax Caiyun Liu, Xiaoxue Luo, Jie Xiong 3/18/2026

High-Fidelity Compression of Seismic Velocity Models via SIREN Auto-Decoders

Neural compression framework using SIREN auto-decoders for high-fidelity compression of multi-structural seismic velocity models.

Ax Yu Hao, Qiuyu Wang, Cheng Yang, Yawen Li, Zhiqiang Zhang, Chuan Shi 3/18/2026

GNNVerifier: Graph-based Verifier for LLM Task Planning

Graph-based verifier for LLM task planning that identifies and corrects hallucinations and flaws in agent-generated plans.

Ax Sumedha Chugh, Ranjitha Prasad, Nazreen Shah 3/18/2026

Informative Perturbation Selection for Uncertainty-Aware Post-hoc Explanations

Post-hoc model-agnostic explanation method using informative perturbation selection for uncertainty-aware interpretability of black-box ML models.

Ax Hideaki Iiduka 3/18/2026

Muon Converges under Heavy-Tailed Noise: Nonconvex H\"{o}lder-Smooth Empirical Risk Minimization

Convergence analysis of Muon optimizer under heavy-tailed noise for nonconvex optimization in large-scale deep neural network training.

Ax Gal Dalal, Assaf Hallak, Gal Chechik, Yftah Ziser 3/18/2026

More Test-Time Compute Can Hurt: Overestimation Bias in LLM Beam Search

Analysis showing wider beam search in LLMs can degrade output quality due to overestimation bias in noisy scorer outputs, with theoretical grounding.

Ax Seth Karten, Jake Grigsby, Tersoo Upaa Jr, Junik Bae, Seonghun Hong, Hyunyoung Jeong, Jaeyoon Jung, Kun Kerdthaisong, Gyungbo Kim, Hyeokgi Kim, Yujin Kim, Eunju Kwon, Dongyu Liu, Patrick Mariglia, Sangyeon Park, Benedikt Schink, Xianwei Shi, Anthony Sistilli, Joseph Twin, Arian Urdu, Matin Urdu, Qiao Wang, Ling Wu, Wenli Zhang, Kunsheng Zhou, Stephanie Milani, Kiran Vodrahalli, Amy Zhang, Fei Fang, Yuke Zhu, Chi Jin 3/18/2026

The PokeAgent Challenge: Competitive and Long-Context Learning at Scale

Large-scale benchmark for AI agents combining partial observability, game-theoretic reasoning, and long-horizon planning in Pokemon battle environment.

Ax Vasiliy A. Es'kin, Egor V. Ivanov 3/18/2026

Physics-Informed Neural Systems for the Simulation of EUV Electromagnetic Wave Diffraction from a Lithography Mask

Physics-informed neural networks and neural operators for simulating EUV electromagnetic wave diffraction in lithography mask applications.

Ax Qizhi Chen, Delin Qu, Junli Liu, Yiwen Tang, Haoming Song, Dong Wang, Yuan Yuan, Bin Zhao 3/18/2026

FreeGaussian: Annotation-free Control of Articulated Objects via 3D Gaussian Splats with Flow Derivatives

Annotation-free method for reconstructing controllable 3D Gaussian splats of articulated objects from monocular video using flow derivatives.

Ax Ziqi Gao, Weikai Huang, Jieyu Zhang, Aniruddha Kembhavi, Ranjay Krishna 3/18/2026

Generate Any Scene: Scene Graph Driven Data Synthesis for Visual Generation Training

Data synthesis engine using scene graphs to improve compositional generalization and semantic alignment in text-to-image generation models.

Ax Xiao Zhu, Chenmien Tan, Pinzhen Chen, Rico Sennrich, Huiming Wang, Yanlin Zhang, Hanxu Hu 3/18/2026

CHARM: Calibrating Reward Models With Chatbot Arena Scores

CHARM method calibrating reward models using Chatbot Arena scores to mitigate model preference bias, improving alignment of LLMs through RLHF.

Ax Wei-Bang Jiang, Xi Fu, Yi Ding, Cuntai Guan 3/18/2026

Towards Robust Multimodal Physiological Foundation Models: Handling Arbitrary Missing Modalities

PhysioOmni foundation model for multimodal physiological signals handling arbitrary missing modalities across EEG, ECG, EOG, EMG for healthcare and brain-computer interfaces.

Ax Mathew J. Koretsky, Maya Willey, Owen Bianchi, Chelsea X. Alvarado, Tanay Nayak, Nicole Kuznetsov, Sungwon Kim, Mike A. Nalls, Daniel Khashabi, Faraz Faghri 3/18/2026

BiomedSQL: Text-to-SQL for Scientific Reasoning on Biomedical Knowledge Bases

BiomedSQL benchmark for text-to-SQL generation requiring scientific reasoning over biomedical knowledge bases, evaluating LLM capability for complex analytical tasks.

Ax Payel Bhattacharjee, Fengwei Tian, Geoffrey D. Rubin, Joseph Y. Lo, Nirav Merchant, Heidi Hanson, John Gounley, Ravi Tandon 3/18/2026

Learning to Diagnose Privately: DP-Powered LLMs for Radiology Report Classification

DP-Powered LLMs for privacy-preserving radiology report classification, enabling differential privacy in healthcare diagnosis and abnormality classification workflows.

Ax Hyunjong Ok, Jaeho Lee 3/18/2026

TempCore: Are Video QA Benchmarks Temporally Grounded? A Frame Selection Sensitivity Analysis and Benchmark

TempCore benchmark analyzing whether video QA models genuinely require temporal frame selection, introducing Frame Selection Sensitivity metric for VLM diagnostic evaluation.

Ax Federica Uccello, Simin Nadjm-Tehrani 3/18/2026

Feature Attribution in 5G Intrusion Detection: A Statistical vs. Logic-Based Comparison

Comparison of statistical and logic-based XAI techniques for interpreting ML security alerts in 5G intrusion detection systems, enabling actionable incident response.

Ax Jewon Lee, Wooksu Shin, Seungmin Yang, Ki-Ung Song, DongUk Lim, Jaeyeon Kim, Tae-Ho Kim, Bo-Kyeong Kim 3/18/2026

ERGO: Efficient High-Resolution Visual Understanding for Vision-Language Models

ERGO framework for efficient high-resolution image processing in vision-language models using coarse-to-fine reasoning pipeline with two-stage visual token reduction.

Ax Sumanth Varambally, Thomas Voice, Yanchao Sun, Zhifeng Chen, Rose Yu, Ke Ye 3/18/2026

Hilbert: Recursively Building Formal Proofs with Informal Reasoning

Hilbert system recursively building formal proofs by combining informal LLM reasoning with Lean 4 verification, bridging gap between mathematical reasoning and formal proof generation.

Ax Sumanth Varambally, Marshall Fisher, Jas Thakker, Yiwei Chen, Zhirui Xia, Yasaman Jafari, Ruijia Niu, Manas Jain, Veeramakali Vignesh Manivannan, Zachary Novack, Luyu Han, Srikar Eranky, Salva R\"uhling Cachay, Taylor Berg-Kirkpatrick, Duncan Watson-Parris, Yi-An Ma, Rose Yu 3/18/2026

Zephyrus: An Agentic Framework for Weather Science

Zephyrus framework combining foundation models for weather forecasting with LLM reasoning to enable language-based scientific workflows on meteorological datasets.

Ax L\'eo Boisvert, Abhay Puri, Chandra Kiran Reddy Evuru, Nazanin Sepahvand, Nicolas Chapados, Quentin Cappart, Alexandre Lacoste, Krishnamurthy Dj Dvijotham, Alexandre Drouin 3/18/2026

Malice in Agentland: Down the Rabbit Hole of Backdoors in the AI Supply Chain

Security research on backdoor attacks in AI agent supply chains through poisoned interaction data collection, formalizing threat models for finetuned web browsing and tool-use agents.

Ax Daichi Nagai, Ryugo Morita, Shunsuke Kitada, Hitoshi Iyatomi 3/18/2026

TAUE: Training-free Noise Transplant and Cultivation Diffusion Model

Training-free diffusion model enabling layer-wise control in text-to-image generation through noise transplantation without fine-tuning or large datasets.

Ax Po-Heng Chou, Chiapin Wang, Kuan-Hao Chen, Wei-Chen Hsiao 3/18/2026

DRL-Based Beam Positioning for LEO Satellite Constellations with Weighted Least Squares

Deep Q-Network learns satellite weighting for CSI-free multi-satellite positioning in LEO constellations combined with weighted least squares estimation.

Ax Dashti A. Ali, Aras T. Asaad, Jacob J. Peoples, Ahmad Bashir Barekzai, Camila Vilela, Hala Khasawneh, Jayasree Chakraborty, Jo\~ao Miranda, Mohammad Hamghalam, Natalie Gangai, Natally Horvat, Richard K. G. Do, Alice C. Wei, Amber L. Simpson 3/18/2026

A Novel Patch-Based TDA Approach for Computed Tomography Imaging

Topological data analysis patch-based approach for CT imaging feature extraction improving ML model performance on medical diagnosis tasks.

Ax Danxu Liu, Di Wang, Hebaixu Wang, Haoyang Chen, Wentao Jiang, Yilin Cheng, Haonan Guo, Wei Cui, Jing Zhang 3/18/2026

SARMAE: Masked Autoencoder for SAR Representation Learning

Noise-aware masked autoencoder for self-supervised SAR satellite imagery representation learning addressing data scarcity and speckle noise challenges.

Ax Nuoya Xiong, Yuhang Zhou, Hanqing Zeng, Zhaorun Chen, Furong Huang, Shuchao Bi, Lizhu Zhang, Zhuokai Zhao 3/18/2026

Token-Level LLM Collaboration via FusionRoute

FusionRoute enables token-level collaboration between specialized and general-purpose LLMs via dynamic routing, improving efficiency and domain performance.

Ax Kai Wittenmayer, Sukrut Rao, Amin Parchami-Araghi, Bernt Schiele, Jonas Fischer 3/18/2026

CFM: Language-aligned Concept Foundation Model for Vision

Language-aligned concept foundation model decomposing vision representations into human-interpretable concepts with spatial grounding across diverse tasks.

Ax Saeed Khaki, Ashudeep Singh, Nima Safaei, Kamal Ginotra 3/18/2026

VisTIRA: Closing the Image-Text Modality Gap in Visual Math Reasoning via Structured Tool Integration

VisTIRA addresses vision-language model performance gap on visual math problems through structured tool integration and dense formula/layout handling.

Ax William Lugoloobi, Thomas Foster, William Bankes, Chris Russell 3/18/2026

LLMs Encode Their Failures: Predicting Success from Pre-Generation Activations

Linear probes on LLM pre-generation activations predict task success before inference, enabling efficient routing and reducing extended reasoning compute costs.

Ax Shan Yang, Yang Liu 3/18/2026

Descent-Guided Policy Gradient for Scalable Cooperative Multi-Agent Learning

Addresses cross-agent noise in multi-agent reinforcement learning through descent-guided policy gradients, reducing sample complexity from O(N/ε) to improved bounds.

Ax Jihoon Jeong 3/18/2026

Model Medicine: A Clinical Framework for Understanding, Diagnosing, and Treating AI Models

Proposes Model Medicine framework for understanding, diagnosing, and treating disorders in AI models using clinical methodology analogies for interpretability.

Ax Mridankan Mandal 3/18/2026

Fusion Complexity Inversion: Why Simpler Cross View Modules Outperform SSMs and Cross View Attention Transformers for Pasture Biomass Regression

Evaluates vision foundation models adapted for pasture biomass regression, comparing SSMs, transformers, and simpler cross-view modules on agricultural imagery.

Ax Pranjal Rawat 3/18/2026

A Survey of Reinforcement Learning For Economics

Survey introducing reinforcement learning methods to economists for solving high-dimensional dynamic programming problems that resist dimensionality reduction.

Ax Yuanyuan Sun, Timothy Parker, Lara Gierschmann, Sana Shams, Teo Canmetin, Mathieu Duteil, Rokas Gipi\v{s}kis, Ze Shen Chin 3/18/2026

Defining AI Models and AI Systems: A Framework to Resolve the Boundary Problem

Framework defining AI models and AI systems through systematic review of 896 academic papers and 80+ regulatory documents to resolve boundary problem in AI regulation.

Ax Ali Raza, Gurang Gupta, Nikolay Matyunin, Jibesh Patra 3/18/2026

Amnesia: Adversarial Semantic Layer Specific Activation Steering in Large Language Models

Red-teaming study on adversarial activation steering techniques to compromise LLM responses, examining safety risks from semantic layer manipulation.

Ax Linus Folkerts, Will Payne, Simon Inman, Philippos Giavridis, Joe Skinner, Sam Deverett, James Aung, Ekin Zorer, Michael Schmatz, Mahmoud Ghanem, John Wilkinson, Alan Steer, Vy Hong, Jessica Wang 3/18/2026

Measuring AI Agents' Progress on Multi-Step Cyber Attack Scenarios

Evaluates autonomous cyber-attack capabilities of frontier AI models on multi-step attack scenarios, comparing seven models over 18 months at varying inference compute budgets.

Ax Tianhao Qian, Zhuoxuan Li, Jinde Cao, Xinli Shi, Leszek Rutkowski 3/18/2026

Alternating Gradient Flow Utility: A Unified Metric for Structural Pruning and Dynamic Routing in Deep Networks

Gradient flow utility metric for structural pruning and dynamic routing in deep networks, addressing magnitude bias in weight-based heuristics.

Ax Jeffrey Hu, Rongzhi Dong, Ying Feng, Ming Hu, Jianjun Hu 3/18/2026

Foundation-Model Surrogates Enable Data-Efficient Active Learning for Materials Discovery

Foundation models as surrogates for active learning in materials discovery, reducing costly synthesis cycles through better uncertainty estimation.

Ax Hamza Mooraj, George Pantazopoulos, Alessandro Suglia 3/18/2026

AgriPath: A Systematic Exploration of Architectural Trade-offs for Crop Disease Classification

Empirical comparison of CNNs, contrastive VLMs, and generative VLMs for crop disease classification across diverse conditions.

Ax Mahdi Naseri, Zhou Wang 3/18/2026

SHAMISA: SHAped Modeling of Implicit Structural Associations for Self-supervised No-Reference Image Quality Assessment

SHAMISA framework for no-reference image quality assessment using self-supervised learning on unlabeled distorted images.

Ax Sudip Laudari, Sang Hun Baek 3/18/2026

Topology-Preserving Data Augmentation for Ring-Type Polygon Annotations

Data augmentation method for ring-type polygon annotations in floorplan analysis preserving topology during geometric transformations.

Ax Yifei Li, Kang Wu, Wenming Wu, Xiao-Ming Fu 3/18/2026

Masked BRep Autoencoder via Hierarchical Graph Transformer

Self-supervised learning framework using masked BRep autoencoder with hierarchical graph transformer for CAD model representation learning.

Ax Timo Hei{\ss}, Coco B\"ogel, Bernd Bischl, Giuseppe Casalicchio 3/18/2026

Analyzing Error Sources in Global Feature Effect Estimation

Research analyzing error sources in global feature effects (PD/ALE plots) for black-box model interpretation.

Ax Bo Jiang 3/18/2026

HindSight: Evaluating LLM-Generated Research Ideas via Future Impact

HindSight framework evaluates LLM-generated research ideas using time-split evaluation against future publications and citation impact.

HN mimbojimbo 3/18/2026

GSD 2

GSD 2 is a standalone CLI coding agent built on Pi SDK, evolving from a Claude prompt framework to a full agent with session and context control.