publications

Selected representative publications, grouped by research theme.

For the complete and up-to-date list, please see my Google Scholar profile.

Reinforcement Learning, Reward Modeling & Policy Optimization

  1. ICML
    On GRPO Collapse in Search-R1: The Lazy Likelihood-Displacement Death Spiral
    Wenlong Deng, Yi Li, Boqing Gong, Yangjun Ren, Christos Thrampoulidis, and Xiaoxiao Li
    International Conference on Machine Learning, 2026
  2. ICLR
    Token Hidden Reward: Steering Exploration-Exploitation in Group Relative Deep Reinforcement Learning
    Wenlong Deng, Yangjun Ren, Yi Li, Boqing Gong, Danica J. Sutherland, Xiaoxiao Li, and Christos Thrampoulidis
    International Conference on Learning Representations, 2026
  3. ICML-W
    Directional Alignment Mitigates Reward Hacking in Reinforcement Learning for Language Models
    Wenlong Deng, Jiahe Huang, Kaan Ozkara, Yi Li, Christos Thrampoulidis, Xiaoxiao Li, and Youngsuk Park
    ICML Workshop on Aligning Reinforcement Learning Experimentalists and Theorists (AIWILD), 2026
  4. NeurIPS
    On the Effect of Negative Gradient in Group Relative Deep Reinforcement Optimization
    Wenlong Deng, Yangjun Ren, Muchen Li, Danica J. Sutherland, Xiaoxiao Li, and Christos Thrampoulidis
    Advances in Neural Information Processing Systems, 2025
  5. NeurIPS
    A Reinforcement Learning-based Bidding Strategy for Data Consumers in Auction-based Federated Learning
    Xiaoli Tang, Han Yu, and Xiaoxiao Li
    Advances in Neural Information Processing Systems, 2025
  6. arXiv
    GSS: Gated Subspace Steering for Selective Memorization Mitigation in LLMs
    Xin Zhang, Huifang Shang, and Xiaoxiao Li
    arXiv preprint arXiv:2602.08901, 2026

Agentic Systems, Planning & Optimization

  1. Agent
    When Single-Agent with Skills Replace Multi-Agent Systems and When They Fail
    Xiaoxiao Li
    Agentic Skills Workshop, 2026
  2. ICML
    TextResNet: Decoupling and Routing Optimization Signals in Compound AI Systems via Deep Residual Tuning
    Shuo Huang, Muchen Li, Han Yu, and Xiaoxiao Li
    International Conference on Machine Learning, 2026
  3. ICLR
    Textual Equilibrium Propagation for Deep Compound AI Systems
    Minghui Chen, Wenlong Deng, James Zou, Han Yu, and Xiaoxiao Li
    International Conference on Learning Representations, 2026
  4. arXiv
    Spend Less, Reason Better: Budget-Aware Value Tree Search for LLM Agents
    Yi Li, Wenlong Deng, Jiahe Li, and Xiaoxiao Li
    arXiv preprint arXiv:2603.12634, 2026
  5. arXiv
    AgentArk: Distilling Multi-Agent Intelligence into a Single LLM Agent
    Yao Luo, Yizhu Jin, Wei Yu, Muhan Zhang, Sanjiv Kumar, Xiaoxiao Li, and Jindong Wang
    arXiv preprint arXiv:2602.03955, 2026

Data-Centric AI: Valuation, Curation & Synthetic Data

  1. ACL
    Efficient Forward-Only Data Valuation for Pretrained LLMs and VLMs
    Wenlong Deng, Jing Zhang, Qi Zeng, Christos Thrampoulidis, Boqing Gong, and Xiaoxiao Li
    Annual Meeting of the Association for Computational Linguistics, 2026
  2. ICLR
    GMValuator: Similarity-based Data Valuation for Generative Models
    Jiaxi Yang, Wenlong Deng, Benlin Liu, Yangsibo Huang, James Zou, and Xiaoxiao Li
    International Conference on Learning Representations, 2025
  3. ICML-W
    Enhancing Clinical Multiple-Choice Questions Benchmarks with Knowledge Graph Guided Distractor Generation
    Running Yang, Wenlong Deng, Minghui Chen, Yuan Zhou, and Xiaoxiao Li
    ICML Workshop on Reliable and Responsible Foundation Models (RRFM), 2025
  4. arXiv
    MedReason: Eliciting Factual Medical Reasoning Steps in LLMs via Knowledge Graphs
    Juncheng Wu, Wenlong Deng, Xiaoxiao Li, and  others
    arXiv preprint arXiv:2504.00993, 2025
  5. JAIR
    Differentially Private Neural Tangent Kernels (DP-NTK) for Privacy-Preserving Data Generation
    Yilin Yang, Kamil Adamczewski, Xiaoxiao Li, Danica J. Sutherland, and Mijung Park
    Journal of Artificial Intelligence Research, 2024

Evaluation, Robustness & Verification for Deployment

  1. ICML
    When RAG Hurts: Diagnosing and Mitigating Attention Distraction in Retrieval-Augmented LVLMs
    Beidi Zhao, Wenlong Deng, Xiang Liao, Yi Li, Nafisa Shaikh, Yuchen Nie, and Xiaoxiao Li
    International Conference on Machine Learning, 2026
  2. arXiv
    Verification Mirage: Mapping the Reliability Boundary of Self-Verification in Medical VQA
    Ruinan Jin, Beidi Zhao, Mintong Kang, Qiong Zhang, and Xiaoxiao Li
    arXiv preprint arXiv:2605.10850, 2026
  3. arXiv
    RVCBench: Benchmarking the Robustness of Voice Cloning Across Modern Audio Generation Models
    Xiang Liao, Ruinan Jin, Han Yu, Devang Pandya, and Xiaoxiao Li
    arXiv preprint arXiv:2602.00443, 2026
  4. MICCAI
    DUCX: Decomposing Unfairness in Tool-Using Chest X-ray Agents
    Zikang Xu, Ruinan Jin, and Xiaoxiao Li
    International Conference on Medical Image Computing and Computer-Assisted Intervention, 2026
  5. NeurIPS
    FairMedFM: Fairness Benchmarking for Medical Imaging Foundation Models
    Ruinan Jin, Zikang Xu, Yuan Zhong, Qiongsong Yao, Qi Dou, S. Kevin Zhou, and Xiaoxiao Li
    Advances in Neural Information Processing Systems, 2024
  6. SaTML
    Backdoor Attack on Unpaired Medical Image-Text Foundation Models: A Pilot Study on MedCLIP
    Ruinan Jin, Chun-Yin Huang, Chenyu You, and Xiaoxiao Li
    IEEE Conference on Secure and Trustworthy Machine Learning, 2024

Optimization & Distributed (Large-Scale) Learning

  1. ICLR
    Can Textual Gradient Work in Federated Learning?
    Minghui Chen, Ruinan Jin, Wenlong Deng, Yuanyuan Chen, Zhi Huang, Han Yu, and Xiaoxiao Li
    International Conference on Learning Representations, 2025
  2. ICLR
    DARE the Extreme: Revisiting Delta-Parameter Pruning for Fine-Tuned Models
    Wenlong Deng, Yize Zhao, Vala Vakilian, Minghui Chen, Xiaoxiao Li, and Christos Thrampoulidis
    International Conference on Learning Representations (Spotlight), 2025
  3. AAAI
    Federated Causally Invariant Feature Learning
    Xin Guo, Kai Yu, Lei Cui, Han Yu, and Xiaoxiao Li
    Proceedings of the AAAI Conference on Artificial Intelligence, 2025
  4. NeurIPS
    Class-wise Balancing Data Replay for Federated Class-Incremental Learning
    Zhuang Qi, Yuan-Pei Tang, Lei Meng, Han Yu, Xiaoxiao Li, and Xiangxu Meng
    Advances in Neural Information Processing Systems, 2025
  5. ICML
    FedCal: Achieving Local and Global Calibration in Federated Learning via Aggregated Parameterized Scaler
    Hongyi Peng, Han Yu, Xiaoli Tang, and Xiaoxiao Li
    International Conference on Machine Learning, 2024
  6. ICML
    Overcoming Data and Model Heterogeneities in Decentralized Federated Learning via Synthetic Anchors
    Chun-Yin Huang, Karthik Srinivas, Xin Zhang, and Xiaoxiao Li
    International Conference on Machine Learning, 2024
  7. CVPR
    Unlocking the Potential of Prompt-Tuning in Bridging Generalized and Personalized Federated Learning
    Wenlong Deng, Christos Thrampoulidis, and Xiaoxiao Li
    Conference on Computer Vision and Pattern Recognition, 2024
  8. NeurIPS
    Local Superior Soups: A Catalyst for Model Merging in Cross-Silo Federated Learning
    Minghui Chen, Meirui Jiang, Xin Zhang, Qi Dou, Zehua Wang, and Xiaoxiao Li
    Advances in Neural Information Processing Systems, 2024
  9. NeurIPS
    Federated Model Heterogeneous Matryoshka Representation Learning
    Liping Yi, Han Yu, Chao Ren, Gang Wang, Xiaoguang Liu, and Xiaoxiao Li
    Advances in Neural Information Processing Systems, 2024

Vision and Multimodal Perception

  1. MICCAI
    LoFi: Location-Aware Fine-Grained Representation Learning for Chest X-ray
    Myeongkyun Kang, Yanting Yang, and Xiaoxiao Li
    International Conference on Medical Image Computing and Computer-Assisted Intervention, 2026
  2. ICLR
    S4M: S4 for Multivariate Time Series Forecasting with Missing Values
    Jing Peng, Meiqi Yang, Qiong Zhang, and Xiaoxiao Li
    International Conference on Learning Representations, 2025
  3. ICML
    Learning High-Order Relationships of Brain Regions
    Weikang Qiu, Huangrui Chu, Selena Wang, Haolan Zuo, Xiaoxiao Li, Yize Zhao, and Rex Ying
    International Conference on Machine Learning, 2024
  4. MedIA
    MMGPL: Multimodal Medical Data Analysis with Graph Prompt Learning
    Liang Peng, Songyue Cai, Zongqian Wu, Huifang Shang, Xiaofeng Zhu, and Xiaoxiao Li
    Medical Image Analysis, 2024
  5. ACM MM
    A Simple and Provable Approach for Learning on Noisy Labeled Multi-modal Medical Images
    Nan Wang, Zonglin Di, Hui He, Qingchao Jiang, and Xiaoxiao Li
    ACM International Conference on Multimedia, 2024
  6. MICCAI
    Debiased Noise Editing on Foundation Models for Fair Medical Image Classification
    Ruinan Jin, Wenlong Deng, Minghui Chen, and Xiaoxiao Li
    International Conference on Medical Image Computing and Computer-Assisted Intervention, 2024

Biomedical & Scientific Discovery

  1. Nat. Methods
    BUDDY: molecular formula discovery via bottom-up MS/MS interrogation
    Shipei Xing, Sam Shen, Banghua Xu, Xiaoxiao Li, and Tao Huan
    Nature Methods, 2023
  2. MedIA
    PTCMIL: Multiple Instance Learning via Prompt Token Clustering for Whole Slide Image Analysis
    Beidi Zhao, SangMook Kim, Hao Chen, Chen Zhou, Zu-hua Gao, Gang Wang, and Xiaoxiao Li
    Medical Image Analysis, 2026