Ax Xiangyu Yin, Ming Du, Junjing Deng, Zhi Yang, Yimo Han, Yi Jiang 3/9/2026

Autonomous Algorithm Discovery for Ptychography via Evolutionary LLM Reasoning

LLM-driven framework autonomously discovers and evolves regularization algorithms for ptychography imaging using evolutionary mechanisms and code generation.

Ax Medina Andresel, Cristinel Mateis, Dejan Nickovic, Spyridon Kounoupidis, Panagiotis Katsaros, Stavros Tripakis 3/9/2026

LTLGuard: Formalizing LTL Specifications with Compact Language Models and Lightweight Symbolic Reasoning

LTLGuard uses compact LLMs with symbolic reasoning to translate natural language requirements into formal LTL specifications.

Ax Ved Sriraman, Adam Block 3/9/2026

Revisiting the (Sub)Optimality of Best-of-N for Inference-Time Alignment

Theoretical analysis of Best-of-N sampling for inference-time LLM alignment, examining optimality and reward hacking vulnerabilities.

Ax Mykola Pinchuk 3/9/2026

TML-Bench: Benchmark for Data Science Agents on Tabular ML Tasks

TML-Bench evaluates 10 open-source LLMs as autonomous data science agents on Kaggle tasks with time constraints and repeated runs.

Ax Levy Chaves, Chao Zhou, Rebekka Burkholz, Eduardo Valle, Sandra Avila 3/9/2026

Bridging Domains through Subspace-Aware Model Merging

Subspace-aware model merging technique for domain generalization across multiple task-specific fine-tuned models.

Ax Jinman Wu, Yi Xie, Shiqian Zhao, Xiaofeng Chen 3/9/2026

Depth Charge: Jailbreak Large Language Models from Deep Safety Attention Heads

Depth Charge jailbreak attack targets deep safety attention heads in open-source LLMs through internal model components.

Ax Jinman Wu, Yi Xie, Shen Lin, Shiqian Zhao, Xiaofeng Chen 3/9/2026

Knowing without Acting: The Disentangled Geometry of Safety Mechanisms in Large Language Models

Disentangled Safety Hypothesis reveals LLMs separate harmfulness recognition from refusal behavior, explaining jailbreak vulnerability.

Ax Samah Fodeh, Linhai Ma, Ganesh Puthiaraju, Srivani Talakokkul, Afshan Khan, Ashley Hagaman, Sarah Lowe, Aimee Roundtree 3/9/2026

PVminerLLM: Structured Extraction of Patient Voice from Patient-Generated Text using Large Language Models

PVminerLLM extracts structured patient voice data from patient-generated text using LLMs for healthcare outcomes research.

Ax Ruixuan Sun, Matthew Zent, Minzhu Zhao, Thanmayee Boyapati, Xinyi Li, Joseph A. Konstan 3/9/2026

Balancing Domestic and Global Perspectives: Evaluating Dual-Calibration and LLM-Generated Nudges for Diverse News Recommendation

Study of LLM-generated nudges and dual-calibration for diverse news recommendation in real-user testing with 120 readers.

Ax Donghoon Han, Eunhwan Park, Seunghyeon Seo 3/9/2026

Visual Words Meet BM25: Sparse Auto-Encoder Visual Word Scoring for Image Retrieval

BM25-V applies BM25 scoring to sparse visual-word activations from SAE on Vision Transformer features for interpretable image retrieval.

Ax Xisen Jin, Michael Duan, Qin Lin, Aaron Chan, Zhenglun Chen, Junyi Du, Xiang Ren 3/9/2026

Proof-of-Guardrail in AI Agents and What (Not) to Trust from It

Proof-of-guardrail cryptographic system enabling AI agent developers to prove safety measures are enforced using open-source guardrails.

Ax Haoran Qiu, Gohar Irfan Chaudhry, Chaojie Zhang, \'I\~nigo Goiri, Esha Choukse, Rodrigo Fonseca, Ricardo Bianchini 3/9/2026

StreamWise: Serving Multi-Modal Generation in Real-Time at Scale

StreamWise system for real-time serving of multi-modal generative models at scale with efficient model coordination.

Ax Shira Gur-Arieh, Angelina Wang, Sina Fazelpour 3/9/2026

Ambiguity Collapse by LLMs: A Taxonomy of Epistemic Risks

Taxonomy of epistemic risks from LLMs collapsing ambiguous concepts in content moderation, hiring, and constitutional AI alignment.

Ax Salim Khazem 3/9/2026

Margin and Consistency Supervision for Calibrated and Robust Vision Models

MaCS regularization framework for improving calibration and robustness of deep vision classifiers under distribution shift.

Ax Srishti Palani, Vidya Setlur 3/9/2026

Lexara: A User-Centered Toolkit for Evaluating Large Language Models for Conversational Visual Analytics

Lexara toolkit for developer evaluation of LLMs in conversational visual analytics, designed with input from 22 CVA developers and 16 end-users.

Ax Anushka Debnath, Stephen Cranefield, Bastin Tony Roy Savarimuthu, Emiliano Lorini 3/9/2026

Evaluating LLM Alignment With Human Trust Models

White-box analysis of how LLMs internally represent and reason about trust using contrastive prompting on gpt-j-6B.

Ax Niful Islam, Md. Rayhan Ahmed, Nur Mohammad Fahad, Salekul Islam, A. K. M. Muzahidul Islam, Saddam Mukta, Swakkhar Shatabda 3/9/2026

Remote Sensing Image Classification Using Deep Ensemble Learning

Ensemble learning approach combining CNNs and Vision Transformers for remote sensing image classification.

Ax Qian Da, Yijiang Chen, Min Ju, Zheyi Ji, Albert Zhou, Wenwen Wang, Matthew A Abikenari, Philip Chikontwe, Guillaume Larghero, Bowen Chen, Peter Neiglinger, Dingrong Zhong, Shuhao Wang, Wei Xu, Drew Williamson, German Corredor, Sen Yang, Le Lu, Xiao Han, Kun-Hsing Yu, Jun-zhou Huang, Laura Barisoni, Geert Litjens, Anant Madabhushi, Lifeng Zhu, Chaofu Wang, Junhan Zhao, Weiguo Hu 3/9/2026

Computational Pathology in the Era of Emerging Foundation and Agentic AI -- International Expert Perspectives on Clinical Integration and Translational Readiness

Survey of foundation models and AI agents in computational pathology, examining clinical integration challenges and real-world adoption barriers.

Ax Junjie Li, Xinrui Guo, Yuhao Wu, Roy Ka-Wei Lee, Hongzhi Li, Yutao Xie 3/9/2026

Lost in Stories: Consistency Bugs in Long Story Generation by LLMs

arXiv paper analyzing consistency bugs in long-form story generation by LLMs, introducing benchmark for narrative consistency.

Ax Xuan Li, Zhanke Zhou, Zongze Li, Jiangchao Yao, Yu Rong, Lu Zhang, Bo Han 3/9/2026

Reference-guided Policy Optimization for Molecular Optimization via LLM Reasoning

arXiv paper on using LLM reasoning with reference-guided policy optimization for molecular structure optimization.

Ax Tao Zhang, Rui Ma, Shuotao Xu, Peng Cheng, Yongqiang Xiong 3/9/2026

LUMINA: LLM-Guided GPU Architecture Exploration via Bottleneck Analysis

arXiv paper on LUMINA using LLM reasoning to guide GPU architecture design space exploration for LLM inference.

Ax Yuxin Xie, Yuming Chen, Yishan Yang, Yi Zhou, Tao Zhou, Zhen Zhao, Jiacheng Liu, Huazhu Fu 3/9/2026

CORE-Seg: Reasoning-Driven Segmentation for Complex Lesions via Reinforcement Learning

arXiv paper on CORE-Seg using reinforcement learning with MLLMs for complex medical lesion segmentation.

Ax Feiran Li, Qianqian Xu, Shilong Bao, Zhiyong Yang, Xilin Zhao, Xiaochun Cao, Qingming Huang 3/9/2026

BlackMirror: Black-Box Backdoor Detection for Text-to-Image Models via Instruction-Response Deviation

arXiv paper on BlackMirror framework for detecting backdoored text-to-image models in black-box settings.

Ax Sen Fang, Yalin Feng, Yanxin Zhang, Dimitris N. Metaxas 3/9/2026

RAC: Rectified Flow Auto Coder

arXiv paper on RAC, a Rectified Flow Auto Coder replacing VAE with multi-step decoding and bidirectional inference.

Ax Nikita Soni, Dhruv Vijay Kunjadiya, Pratham Piyush Shah, Dikshya Mohanty, H. Andrew Schwartz, Niranjan Balasubramanian 3/9/2026

Addressing the Ecological Fallacy in Larger LMs with Human Context

arXiv paper addressing ecological fallacy in larger LMs by modeling author language context to improve performance.

Ax Arun Joshi 3/9/2026

XAI for Coding Agent Failures: Transforming Raw Execution Traces into Actionable Insights

arXiv paper on XAI approach transforming execution traces from LLM-based coding agents into actionable debugging insights.

Ax Jialuo He, Huangxun Chen 3/9/2026

Energy-Driven Adaptive Visual Token Pruning for Efficient Vision-Language Models

arXiv paper on E-AdaPrune, an energy-driven adaptive token pruning framework for efficient Vision-Language Models.

Ax Nikita Soni, August H{\aa}kan Nilsson, Syeda Mahwish, Vasudha Varadarajan, H. Andrew Schwartz, Ryan L. Boyd 3/9/2026

Who We Are, Where We Are: Mental Health at the Intersection of Person, Situation, and Large Language Models

arXiv paper using LLMs to predict well-being and identify self-states in longitudinal social media data based on psychological theory.

Ax Junming Liu, Yusen Zhang, Rongchao Zhang, Wenkai Zhu, Tian Wu 3/9/2026

Domain-Adaptive Model Merging across Disconnected Modes

arXiv paper on DMM framework for data-free model merging across domains while preserving privacy and avoiding retraining.

Ax Siyuan Yang, Jun Liu, Hao Cheng, Chong Wang, Shijian Lu, Hedvig Kjellstrom, Weisi Lin, Alex C. Kot 3/9/2026

Skeleton-to-Image Encoding: Enabling Skeleton Representation Learning via Vision-Pretrained Models

arXiv paper on enabling skeleton representation learning by encoding 3D skeleton data for use with vision-pretrained models.

Ax Jiayang Sun, Zixin Guo, Min Cao, Guibo Zhu, Jorma Laaksonen 3/9/2026

Imagine How To Change: Explicit Procedure Modeling for Change Captioning

arXiv paper on ProCap framework for change captioning using dynamic procedure modeling instead of static image comparison.

Ax Zhouchonghao Wu, Raymond Song, Vedant Mundheda, Luis E. Navarro-Serment, Christof Schoenborn, Jeff Schneider 3/9/2026

TADPO: Reinforcement Learning Goes Off-road

Reinforcement learning approach for off-road autonomous driving addressing long-horizon planning and adaptable control in unmapped, variable terrain.

Ax Zhi Lei, Chenxi Liu, Hao Miao, Wanghui Qiu, Bin Yang, Chenjuan Guo 3/9/2026

MM-ISTS: Cooperating Irregularly Sampled Time Series Forecasting with Multimodal Vision-Text LLMs

Multimodal framework leveraging vision-text LLMs for irregularly sampled time series forecasting with contextual semantics and temporal patterns.

Ax Ninghao Zhang, Bin Zhu, Shijie Zhou, Jingjing Chen 3/9/2026

Restoring Linguistic Grounding in VLA Models via Train-Free Attention Recalibration

Train-free attention recalibration method improving Vision-Language-Action model reliability on out-of-distribution language instructions for robotic manipulation.

Ax Minjong Cheon 3/9/2026

Demystifying KAN for Vision Tasks: The RepKAN Approach

RepKAN architecture combining CNNs with Kolmogorov-Arnold Networks for interpretable remote sensing image classification via spectral fingerprinting.

Ax Yang Liu, Jinxuan Cai, Yishen Li, Qi Meng, Zedi Liu, Xin Li, Chen Qian, Chuan Shi, Cheng Yang 3/9/2026

MASFactory: A Graph-centric Framework for Orchestrating LLM-Based Multi-Agent Systems with Vibe Graphing

Framework for orchestrating LLM-based multi-agent systems using graph-centric workflow modeling and vibe graphing to reduce implementation complexity.

Ax Maik Larooij 3/9/2026

Sensitivity-Aware Retrieval-Augmented Intent Clarification

Retrieval-augmented approach for intent clarification in conversational search systems informed by sensitivity analysis in exploratory search.

Ax Nikos Theodoridis, Reenu Mohandas, Ganesh Sistu, Anthony Scanlan, Ciar\'an Eising, Tim Brophy 3/9/2026

Probing Visual Concepts in Lightweight Vision-Language Models for Automated Driving

Examines intermediate activations in lightweight vision-language models to understand failure modes on visual reasoning tasks relevant to autonomous driving.

Ax Jonas Kubesch, Lena Huber, Clemens Havas 3/9/2026

Evaluating Austrian A-Level German Essays with Large Language Models for Automated Essay Scoring

Investigates open-weight LLMs for automated essay scoring on Austrian A-Level German essays to reduce grading workload and subjective bias.

Ax Xudong Wang, Jiahua Dong, Baichen Liu, Qi Lyu, Lianqing Liu, Zhi Han 3/9/2026

Lifelong Embodied Navigation Learning

Formalizes lifelong embodied navigation learning where LLM-powered agents continually acquire navigation skills across multiple scenes while avoiding catastrophic forgetting.

Ax Xi Wang, Mengdie Zhuang, Jiqun Liu 3/9/2026

Experiences Build Characters: The Linguistic Origins and Functional Impact of LLM Personality

Studies how domain-specific continued pre-training shapes LLM personality traits and influences problem-solving performance through diverse behavioral tendencies.

Ax Puneet Mathur, Branislav Kveton, Subhojyoti Mukherjee, Viet Dac Lai 3/9/2026

Partial Policy Gradients for RL in LLMs

Proposes partial policy gradients approach for reinforcement learning in LLMs by optimizing subsets of future rewards for more reliable policy learning.

Ax Bohai Gu, Taiyi Wu, Dazhao Du, Jian Liu, Shuai Yang, Xiaotong Zhao, Alan Zhao, Song Guo 3/9/2026

Place-it-R1: Unlocking Environment-aware Reasoning Potential of MLLM for Video Object Insertion

MLLM framework for physically consistent video object insertion using environment-aware reasoning from multimodal large language models.

Ax Bj\"orn van Zwol 3/9/2026

Predictive Coding Graphs are a Superset of Feedforward Neural Networks

Proves predictive coding graphs are a mathematical superset of feedforward neural networks, bridging neuroscience-inspired models with contemporary ML.

Ax Rohit Saxena, Alessandro Suglia, Pasquale Minervini 3/9/2026

VLM-RobustBench: A Comprehensive Benchmark for Robustness of Vision-Language Models

VLM-RobustBench: comprehensive benchmark evaluating robustness of vision-language models across 133 corrupted settings and perturbation types.

Ax Ajinkya Kulkarni, Sandipana Dowerah, Atharva Kulkarni, Tanel Alum\"ae, Mathew Magimai Doss 3/9/2026

Do Compact SSL Backbones Matter for Audio Deepfake Detection? A Controlled Study with RAPTOR

RAPTOR: controlled study of compact SSL backbones for audio deepfake detection using pairwise-gated transformer fusion.

Ax Zikai Zhou, Muyao Wang, Shitong Shao, Lichen Bai, Haoyi Xiong, Bo Han, Zeke Xie 3/9/2026

Reflective Flow Sampling Enhancement

Inference-time enhancement strategies for flow-matching text-to-image diffusion models to improve generation quality.

Ax Mohammed Baharoon, Thibault Heintz, Siavash Raissi, Mahmoud Alabbad, Mona Alhammad, Hassan AlOmaish, Sung Eun Kim, Oishi Banerjee, Pranav Rajpurkar 3/9/2026

CRIMSON: A Clinically-Grounded LLM-Based Metric for Generative Radiology Report Evaluation

CRIMSON: clinically-grounded LLM-based evaluation metric for chest X-ray report generation assessing diagnostic correctness and safety.

Ax Hoseong Ahn, Jeongyun Chae, Yoonji Park, Kyuhong Shim 3/9/2026

Whisper-CD: Accurate Long-Form Speech Recognition using Multi-Negative Contrastive Decoding

Whisper-CD: training-free contrastive decoding framework reducing hallucinations in long-form speech recognition.

Ax Naifan Zhang, Ruihan Sun, Jinwei Su, Hengjie Yang, Zhengyuan Pan, Zhaohan Chen, Xiaofan Zhang 3/9/2026

MAPO: Mixed Advantage Policy Optimization for Long-Horizon Multi-Turn Dialogue

MAPO: reinforcement learning approach for optimizing long-horizon multi-turn dialogue policies with outcome-only supervision.