Ax Ning Yang, Hai Lin, Yibo Liu, Baoliang Tian, Guoqing Liu, Haijun Zhang 3/3/2026

Token-Importance Guided Direct Preference Optimization

Token-Importance Guided DPO: Enhanced direct preference optimization for LLM alignment using token-level importance weighting beyond standard DPO.

Ax Yu Wang, Junshu Dai, Yuchen Ying, Hanyang Yuan, Zunlei Feng, Tongya Zheng, Mingli Song 3/3/2026

Adaptive Location Hierarchy Learning for Long-Tailed Mobility Prediction

Adaptive location hierarchy learning method for long-tailed mobility prediction addressing biased predictions in human location forecasting.

Ax Tianhui Liu, Hetian Pang, Xin Zhang, Tianjian Ouyang, Zhiyuan Zhang, Jie Feng, Yong Li, Pan Hui 3/3/2026

CityLens: Evaluating Large Vision-Language Models for Urban Socioeconomic Sensing

CityLens: Benchmark for evaluating large vision-language models' ability to predict socioeconomic indicators from satellite and street view imagery.

Ax Tri Cao, Bennett Lim, Yue Liu, Yuan Sui, Yuexin Li, Shumin Deng, Lin Lu, Nay Oo, Shuicheng Yan, Bryan Hooi 3/3/2026

VPI-Bench: Visual Prompt Injection Attacks for Computer-Use Agents

VPI-Bench: Security research on visual prompt injection attacks against computer-use agents with full system access.

Ax Frederico Metelo, Alexandre Oliveira, Stevo Rackovi\'c, Pedro \'Akos Costa, Cl\'audia Soares 3/3/2026

FAuNO: Semi-Asynchronous Federated Reinforcement Learning Framework for Task Offloading in Edge Systems

FAuNO: Federated asynchronous reinforcement learning framework for task offloading in decentralized edge computing systems.

Ax Mikhail Terekhov, Zhen Ning David Liu, Caglar Gulcehre, Samuel Albanie 3/3/2026

Control Tax: The Price of Keeping AI in Check

Control Tax: Framework for measuring operational and financial costs of integrating AI control mechanisms into agentic AI pipelines for high-stakes applications.

Ax Kihyun Kim, Jiawei Zhang, Asuman Ozdaglar, Pablo A. Parrilo 3/3/2026

Beyond RLHF and NLHF: Population-Proportional Alignment under an Axiomatic Framework

Novel preference learning framework for aligning aggregate opinions proportionally with population distribution rather than majority preference.

Ax Zofia Rudnicka, Janusz Szczepanski, Agnieszka Pregowska 3/3/2026

Accuracy-Efficiency Trade-Offs in Spiking Neural Networks: A Lempel-Ziv Complexity Perspective on Learning Rules

Study of spiking neural networks' accuracy-efficiency trade-offs using Lempel-Ziv complexity to compare unsupervised, supervised, and hybrid learning paradigms.

Ax Akash Gupta, Amos Storkey, Mirella Lapata 3/3/2026

Meta-Adaptive Prompt Distillation for Few-Shot Visual Question Answering

Meta-Adaptive Prompt Distillation for few-shot visual question answering in large multimodal models using optimized in-context learning.

Ax Cong Chen, Omer Karaduman, Xu Kuang 3/3/2026

Behavioral Generative Agents for Energy Operations

Study of generative agents for modeling consumer behavior in energy operations, examining their role in operational decision-making and uncertainty handling.

Ax Shuocun Yang, Huawen Hu, Xuan Liu, Yincheng Yao, Enze Shi, Shu Zhang 3/3/2026

Structured Diversity Control: A Dual-Level Framework for Group-Aware Multi-Agent Coordination

Dual-level framework for controlling behavioral diversity in multi-agent reinforcement learning systems accounting for agent composition structures.

Ax Bo Liu, Leon Guertler, Simon Yu, Zichen Liu, Penghui Qi, Daniel Balcells, Mickel Liu, Cheston Tan, Weiyan Shi, Min Lin, Wee Sun Lee, Natasha Jaques 3/3/2026

SPIRAL: Self-Play on Zero-Sum Games Incentivizes Reasoning via Multi-Agent Multi-Turn Reinforcement Learning

SPIRAL: Self-play reinforcement learning framework where language models play zero-sum games against themselves to develop reasoning without human-curated data.

Ax Shuhao Mei, Yongchao Long, Xiaoyu Xiao, Shan Cao, Xiaobo Han, Shijia Geng, Jinbo Sun, Yuxi Zhou, Shenda Hong 3/3/2026

SpiroLLM: Finetuning Pretrained LLMs to Understand Spirogram Time Series with Clinical Validation in COPD Reporting

SpiroLLM: Fine-tuned LLM for analyzing spirogram time series to predict COPD respiratory disease with clinical validation.

Ax Ashmi Banerjee, Adithi Satish, Fitri Nur Aisyah, Wolfgang W\"orndl, Yashar Deldjoo 3/3/2026

Collab-REC: An LLM-based Agentic Framework for Balancing Recommendations in Tourism

Collab-REC: Multi-agent LLM framework with three specialized agents negotiating tourism recommendations to balance personalization, popularity, and sustainability.

Ax Ao Cheng, Lei Zhang, Guowei He 3/3/2026

Re4: Scientific Computing Agent with Rewriting, Resolution, Review and Revision

Re4: LLM-based agent framework for scientific computing using rewriting-resolution-review-revision chain for complex mathematical and scientific reasoning tasks.

Ax Jonathn Chang, Leonhard Piff, Suvadip Sana, Jasmine X. Li, Lionel Levine 3/3/2026

EigenBench: A Comparative Behavioral Measure of Value Alignment

EigenBench: Black-box method for benchmarking language models' value alignment using constitutional scoring without access to model internals.

Ax Qika Lin, Yifan Zhu, Bin Pu, Ling Huang, Haoran Luo, Jingying Ma, Feng Wu, Kai He, Jiaxing Xu, Zhen Peng, Tianzhe Zhao, Fangzhi Xu, Jian Zhang, Zhonghong Ou, Erik Cambria, Swapnil Mishra, Mengling Feng 3/3/2026

Toward Clinically Explainable AI for Medical Diagnosis: A Foundation Model with Human-Compatible Reasoning via Reinforcement Learning

DeepMedix-R1 foundation model for chest X-ray interpretation generates diagnoses with step-by-step reasoning process for clinical explainability.

Ax Sasi Kiran Gaddipati, Farhana Keya, Gollam Rabby, S\"oren Auer 3/3/2026

AISSISTANT: Human-AI Collaborative Review and Perspective Research Workflows in Data Science

AIssistant: open-source agentic framework for human-AI collaborative generation of scientific review and perspective papers, combining autonomous workflows with human guidance.

Ax Zongru Wu, Rui Mao, Zhiyuan Tian, Pengzhou Cheng, Tianjie Ju, Zheng Wu, Lingzhong Dong, Haiyue Sheng, Zhuosheng Zhang, Gongshen Liu 3/3/2026

See, Think, Act: Teaching Multimodal Agents to Effectively Interact with GUI by Identifying Toggles

Framework teaches multimodal agents to reliably interact with GUIs by identifying toggle controls, addressing key bottleneck in automated interface manipulation.

Ax Miao Li, Alexander Gurung, Irina Saparina, Mirella Lapata 3/3/2026

Who Gets Cited Most? Benchmarking Long-Context Numerical Reasoning on Scientific Articles

SciTrek benchmark probes long-context numerical reasoning in LLMs using full-text scientific articles, testing counting, sorting, aggregating, and comparing across documents.

Ax Dong-Kyum Kim, Minsung Kim, Jea Kwon, Nakyeong Yang, Meeyoung Cha 3/3/2026

Bilinear representation mitigates reversal curse and enables consistent model editing

Research shows bilinear representation in neural networks mitigates the reversal curse and enables consistent model editing, revealing knowledge encoding artifacts rather than fundamental limitations.

Ax Gyubok Lee, Woosog Chay, Heeyoung Kwak, Yeong Hwa Kim, Haanju Yoo, Oksoon Jeong, Meong Hi Son, Edward Choi 3/3/2026

From Conversation to Query Execution: Benchmarking User and Tool Interactions for EHR Database Agents

EHR-ChatQA benchmark evaluates LLM-powered agents for Electronic Health Record database access, addressing query ambiguity and terminology mismatch in clinical workflows.

Ax Zhuoli Yin, Yi Ding, Reem Khir, Hua Cai 3/3/2026

ViTSP: A Vision Language Models Guided Framework for Solving Large-Scale Traveling Salesman Problems

ViTSP framework uses vision language models to solve large-scale Traveling Salesman Problems with improved generalization and scalability over traditional learning-based approaches.

Ax Yuhui Wang, Changjiang Li, Guangke Chen, Jiacheng Liang, Ting Wang 3/3/2026

Reasoning or Retrieval? A Study of Answer Attribution on Large Reasoning Models

Study analyzing answer attribution in large reasoning models, showing inconsistencies between chain-of-thought reasoning and final outputs.

Ax Linhao Luo, Zicheng Zhao, Junnan Liu, Zhangchi Qiu, Junnan Dong, Serge Panev, Chen Gong, Thuy-Trang Vu, Gholamreza Haffari, Dinh Phung, Alan Wee-Chung Liew, Shirui Pan 3/3/2026

G-reasoner: Foundation Models for Unified Reasoning over Graph-structured Knowledge

G-reasoner foundation model enables unified reasoning over graph-structured knowledge, combining LLM capabilities with structured knowledge integration.

Ax Yichi Zhang, Yue Ding, Jingwen Yang, Tianwei Luo, Dongbai Li, Ranjie Duan, Qiang Liu, Hang Su, Yinpeng Dong, Jun Zhu 3/3/2026

Towards Safe Reasoning in Large Reasoning Models via Corrective Intervention

Framework for detecting and correcting harmful content in chain-of-thought reasoning of large reasoning models via corrective interventions.

Ax Xinpeng Wang, Nitish Joshi, Barbara Plank, Rico Angell, He He 3/3/2026

Is It Thinking or Cheating? Detecting Implicit Reward Hacking by Measuring Reasoning Effort

TRACE method detects implicit reward hacking in reasoning models by measuring reasoning effort via truncated CoT analysis.

Ax Adel Javanmard, Baharan Mirzasoleiman, Vahab Mirrokni 3/3/2026

Understanding the Role of Training Data in Test-Time Scaling

Analysis of training data conditions enabling test-time scaling in LLMs; examines when long chains-of-thought emerge during training.

Ax Xu Shen, Song Wang, Zhen Tan, Laura Yao, Xinyu Zhao, Kaidi Xu, Xin Wang, Tianlong Chen 3/3/2026

FaithCoT-Bench: Benchmarking Instance-Level Faithfulness of Chain-of-Thought Reasoning

FaithCoT-Bench benchmarks instance-level faithfulness of chain-of-thought reasoning to assess reliability in high-risk LLM applications.

Ax Yunghwei Lai, Kaiming Liu, Ziyue Wang, Weizhi Ma, Yang Liu 3/3/2026

Doctor-R1: Mastering Clinical Inquiry with Experiential Agentic Reinforcement Learning

Doctor-R1 agent uses experiential agentic reinforcement learning to master both clinical decision-making and strategic patient inquiry skills.

Ax Gang Li, Yan Chen, Ming Lin, Tianbao Yang 3/3/2026

DRPO: Efficient Reasoning via Decoupled Reward Policy Optimization

DRPO optimization technique reduces computational cost of reasoning models by mitigating overthinking and unnecessary lengthy reasoning chains.

Ax Jiaxin Gao, Chen Chen, Yanwen Jia, Xueluan Gong, Kwok-Yan Lam, Qian Wang 3/3/2026

Evaluating and Mitigating LLM-as-a-judge Bias in Communication Systems

Systematic investigation of bias in LLM-as-a-judge systems across 6 models; evaluates fairness issues in autonomous content quality assessment.

Ax Jingcong Liang, Shijun Wan, Xuehai Wu, Yitong Li, Qianglong Chen, Duyu Tang, Siyuan Wang, Zhongyu Wei 3/3/2026

HardcoreLogic: Challenging Large Reasoning Models with Long-tail Logic Puzzle Games

HardcoreLogic benchmark tests large reasoning models on non-canonical logic puzzle variants to evaluate flexible rule application and generalization.

Ax Hanane Nour Moussa, Patrick Queiroz Da Silva, Daniel Adu-Ampratwum, Alyson East, Zitong Lu, Nikki Puccetti, Mingyi Xue, Huan Sun, Bodhisattwa Prasad Majumder, Sachin Kumar 3/3/2026

ScholarEval: Research Idea Evaluation Grounded in Literature

ScholarEval: retrieval-augmented evaluation framework assessing research ideas on soundness and contribution using literature grounding.

Ax Yuanhe Zhang, Ilja Kuzborskij, Jason D. Lee, Chenlei Leng, Fanghui Liu 3/3/2026

DAG-Math: Graph-of-Thought Guided Mathematical Reasoning in LLMs

DAG-Math models chain-of-thought reasoning as rule-based processes on directed acyclic graphs to analyze LLM mathematical problem-solving mechanisms.

Ax Christian Dittrich, Jennifer Flygare Kinne 3/3/2026

The Information-Theoretic Imperative: Compression and the Epistemic Foundations of Intelligence

Theoretical analysis of representation convergence in brains and neural networks through compression efficiency principle framework.

Ax Annan Li, Chufan Wu, Zengle Ge, Yee Hin Chong, Zhinan Hou, Lizhe Cao, Cheng Ju, Jianmin Wu, Huaiming Li, Haobo Zhang, Shenghao Feng, Mo Zhao, Fengzhi Qiu, Rui Yang, Mengmeng Zhang, Wenyi Zhu, Yingying Sun, Quan Sun, Shunhao Yan, Danyu Liu, Dawei Yin, Dou Shen 3/3/2026

The FM Agent

FM Agent: multi-agent framework combining LLM reasoning with large-scale evolutionary search for complex scientific and engineering discovery tasks.

Ax Dirk U. Wulff, Rui Mata 3/3/2026

Addressing Longstanding Challenges in Cognitive Science with Language Models

Survey of language model capabilities and limitations for addressing cognitive science research challenges in integration, formalization, and conceptual clarity.

Ax ChengZhang Yu, YingRu He, Hongyan Cheng, nuo Cheng, Zhixing Liu, Dongxu Mu, Zhangrui Shen Yang Gao, and Zhanpeng Jin 3/3/2026

From Passive to Proactive: A Hierarchical Multi-Agent Framework for Automated Medical Pre-Consultation

Hierarchical multi-agent system for medical triage addressing specialization, department heterogeneity, and real-time responsiveness in healthcare AI.

Ax Chao Wu, Baoheng Li, Mingchen Gao, Yu Tian, Zhenyi Wang 3/3/2026

From Efficiency to Adaptivity: A Deeper Look at Adaptive Reasoning in Large Language Models

Analysis of adaptive reasoning in LLMs showing models apply uniform strategies regardless of task difficulty; proposes methods for adaptive compute allocation.

Ax Ruiyu Wang, Tuan Vinh, Ran Xu, Yuyin Zhou, Jiaying Lu, Carl Yang, Francisco Pasquel 3/3/2026

Knowledge Graph Augmented Large Language Models for Disease Prediction

Knowledge graph-guided chain-of-thought framework improves disease prediction on EHR data by scaffolding reasoning with structured medical knowledge paths.

Ax Elinor Poole-Dayan, Jiayi Wu, Taylor Sorensen, Jiaxin Pei, Michiel A. Bakker 3/3/2026

Benchmarking Overton Pluralism in LLMs

OVERTONBENCH framework measures viewpoint diversity in LLM outputs through set coverage metrics, validated against 1208-person human study across 8 models.

Ax Guang Yang, Tianpei Yang, Jingwen Qiao, Yanqing Wu, Jing Huo, Xingguo Chen, Yang Gao 3/3/2026

Multi-Agent Reinforcement Learning with Communication-Constrained Priors

Multi-agent reinforcement learning framework handling lossy communication via generalized communication-constrained priors for scalable cooperative policy learning.

Ax Ely Hahami, Ishaan Sinha, Lavik Jain, Josh Kaplan, Jon Hahami 3/3/2026

Detecting the Disturbance: A Nuanced View of Introspective Abilities in LLMs

Introspection study on Llama-3.1 using activation steering to test whether LLMs can detect perturbations to internal states; finds artifact in binary detection paradigm.

Ax Haipeng Luo, Huawen Feng, Qingfeng Sun, Can Xu, Kai Zheng, Yufei Wang, Tao Yang, Han Hu, Yansong Tang 3/3/2026

AgentMath: Empowering Mathematical Reasoning for Large Language Models via Tool-Augmented Agent

AgentMath: Agent framework integrating LLM reasoning with code interpreters for accurate mathematical problem-solving, addressing computational limitations of reasoning models.

Ax Mehmet Kurmaz 3/3/2026

AWARE-US: Preference-Aware Infeasibility Resolution in Tool-Calling Agents

AWARE-US: Tool-calling agent system that resolves database query infeasibility by preference-aware repair, handling underspecified and infeasible queries.

Ax Yuxuan Zhou, Fei Huang, Heng Li, Fengyi Wu, Tianyu Wang, Jianwei Zhang, Junyang Lin, Zhi-Qi Cheng 3/3/2026

Overcoming Joint Intractability with Lossless Hierarchical Speculative Decoding

Lossless Hierarchical Speculative Decoding: Improves LLM inference speed via hierarchical verification approach that overcomes joint intractability in token acceptance.

Ax Zibin Meng, Kani Chen 3/3/2026

PsyAgent: Constructing Human-like Agents Based on Psychological Modeling and Contextual Interaction

PsyAgent: Framework for building human-like AI agents using Big Five personality traits and social context conditioning to enable stable dispositions with adaptive behavior.

Ax Shengkai Chen, Zhiguang Cao, Jianan Zhou, Yaoxin Wu, Senthilnath Jayavelu, Zhuoyi Lin, Xiaoli Li, Shili Xiang 3/3/2026

DRAGON: LLM-Driven Decomposition and Reconstruction Agents for Large-Scale Combinatorial Optimization

DRAGON combines LLM-guided decomposition and reconstruction agents for large-scale combinatorial optimization with improved scalability beyond 30 nodes.

Ax Mohammad Pivezhandi, Mahdi Banisharif, Abusayeed Saifullah, Ali Jannesari 3/3/2026

ZeroDVFS: Zero-Shot LLM-Guided Core and Frequency Allocation for Embedded Platforms

Zero-shot LLM-guided framework for dynamic voltage/frequency scaling and core allocation on embedded systems using hierarchical multi-agent RL.