Publications

Preprints and Submitted Papers

1. Why Ghost Outputs Teach: A Kernel-Based Understanding of Subliminal Learning
   Zhe Li, Bicheng Ying, Chaosheng Dong, and Haibo Yang.
   Under Review.

Conference Papers

1. Learning Dynamics of Zeroth-Order Optimization: A Kernel Perspective
   Zhe Li, Bicheng Ying, Zidong Liu, and Haibo Yang.
   International Conference on Machine Learning (ICML), 2026. (acceptance rate: 26.6%)

2. HOSL: Hybrid-Order Split Learning for Memory-Constrained Edge Training
   Aakriti Lnu, Zhe Li, Dandan Liang, Chao Huang, Rui Li, and Haibo Yang.
   International Symposium on Modeling and Optimization in Mobile, Ad hoc, and Wireless Networks (WiOpt), 2026.

3. Converge Faster, Talk Less: Hessian-Informed Federated Zeroth-Order Optimization
   Zhe Li, Bicheng Ying, Zidong Liu, Chaosheng Dong, and Haibo Yang.
   International Conference on Learning Representations (ICLR), 2026. (acceptance rate: 28%)

4. Achieving Extremely Low Communication Overhead in Federated Learning via Zeroth-Order SignSGD
   Zhe Li, Bicheng Ying, Dandan Liang, Zidong Liu, Rui Li, and Haibo Yang.
   Asilomar Conference on Signals, Systems, and Computers, 2025. (Invited Paper)
   

5. Exact and Linear Convergence for Federated Learning under Arbitrary Client Participation is Attainable
   Bicheng Ying, Zhe Li, and Haibo Yang.
   Advances in Neural Information Processing Systems (NeurIPS), 2025. (acceptance rate: 24.52%)

6. Towards Straggler-Resilient Split Federated Learning: An Unbalanced Update Approach
   Dandan Liang, Jianing Zhang, Evan Chen, Zhe Li, Rui Li, and Haibo Yang.
   Advances in Neural Information Processing Systems (NeurIPS), 2025. (acceptance rate: 24.52%)

7. FAST: A Lightweight Mechanism Unleashing Arbitrary Client Participation in Federated Learning
   Zhe Li, Seyedsina Nabavirazavi, Bicheng Ying, Sitharama Iyengar, and Haibo Yang.
   International Joint Conference on Artificial Intelligence (IJCAI), 2025. (acceptance rate: 19.3%)

8. Achieving Dimension-Free Communication in Federated Learning via Zeroth-Order Optimization
   Zhe Li, Bicheng Ying, Zidong Liu, Chaosheng Dong, and Haibo Yang.
   International Conference on Learning Representations (ICLR), 2025. (acceptance rate: 32.08%)

Journal Papers

1. Efficient Fine-Tuning of Large Language Models with Zeroth-Order Model Parallelism
   Zhe Li, Bicheng Ying, Zidong Liu, Julie Huang, and Haibo Yang.
   Transactions on Machine Learning Research (TMLR), 2026.
   

2. Exploring LLMs' Potential for Privacy Leakage Detection in Android App Logs: An Empirical Study
   Zhiyuan Chen, Vanessa Nava-Camal, Tiash Roy, Zhe Li, Yiming Tang, Xueling Zhang, and Haibo Yang.
   IEEE Software, 2026.

Workshop Papers

1. Democratizing LLM Training Across Geo-Distributed Heterogeneous Compute: A Systems Position and Roadmap
   Ziyue Luo, Jiaxuan Cai, Cedric Le Denmat, Srijith Nair, Fatemeh Nourzad, Rohith Krishnan Sudha, Qinhang Wu, Jifan Zhang, Sungjae Lee, Zhe Li, Peiwen Qiu, Siddharth Shah, Rishabh Sharma, Sundararajan Srinivasan, Yinglun Xia, Xue Zheng, Zidong Liu, Bicheng Ying, Kaushik Chowdhury, Gauri Joshi, Yingbin Liang, Rob Nowak, Srini Parthasarathy, Saurav Prakash, Balaraman Ravindran, Sanjay Shakkottai, Ness B. Shroff, Sundar Srinivasan, Haibo Yang, Aylin Yener, and Jia Liu.
   NeurIPS Workshop on Collaborative, Open, and DECentralized training of Foundation Models (CODEC-FM), 2026.
   A shorter version "Toward WAN-Aware LLM Training Across Heterogeneous, Geo-Distributed Sites" has been published at SIGCOMM Workshop on Networks for AI Computing (NAIC) 2026.
   

2. Why Ghost Outputs Teach: Towards a Mechanistic Understanding of Subliminal Learning
   Zhe Li, Bicheng Ying, Chaosheng Dong, and Haibo Yang.
   NeurIPS Workshop on Interpretability as a Science (InterpScience), 2026.
   

3. SGIQ: Sparse-Gradient-Informed Gradient Estimation for Variational Quantum Circuits
   Yujia Cai, Zhe Li, Chaowen Guan, and Haibo Yang.
   NeurIPS Workshop on Secure and Trustworthy Quantum Machine Learning (SaTQuML), 2026.