大模型弱到强泛化机理分析和设计，国自然青年科学基金项目（C类），30万，2026.1-2028.12，负责人

应用为中心的跨域资源调度与数据传输，科技部-国家重点研发计划项目课题，900万，2025.5-2028.4，核心骨干

近年主要发表论文如下：

1. Theoretically Grounded Framework for LLM Watermarking: A Distribution-Adaptive Approach
2. Haiyun He, Yepeng Liu, Ziqiao Wang, Yongyi Mao, and Yuheng Bu
3. NeurIPS 2025
4. 
   
5. CHPO: Constrained Hybrid-action Policy Optimization for Reinforcement Learning
6. Ao Zhou, Jiayi Guan, Li Shen, Fan Lu, Sanqing Qu, Junqiao Zhao, Ziqiao Wang, Ya Wu, and Guang Chen
7. NeurIPS 2025
8. 
   
9. MutualVPR: A Mutual Learning Framework for Resolving Supervision Inconsistencies via Adaptive Clustering
10. Qiwen Gu, Xufei Wang, Junqiao Zhao, Siyue Tao, Tiantian Feng, Ziqiao Wang, and Guang Chen
11. NeurIPS 2025
12. 
    
13. Distributional Information Embedding: A Framework for Multi-bit Watermarking
14. Haiyun He, Yepeng Liu, Ziqiao Wang, Yongyi Mao, and Yuheng Bu
15. APWDSIT 2025
16. 
    
17. Revisiting Weak-to-Strong Generalization in Theory and Practice: Reverse KL vs. Forward KL
18. Wei Yao, Wenkai Yang, Ziqiao Wang, Yankai Lin, and Yong Liu
19. ACL Findings 2025
20. 
    
21. Dynamic Task Vector Grouping for Efficient Multi-Task Prompt Tuning
22. Peiyi Zhang, Richong Zhang, Zhijie Nie, and Ziqiao Wang 
23. ACL Findings 2025
24. 
    
25. Generalization in Federated Learning: A Conditional Mutual Information Framework
26. Ziqiao Wang, Cheng Long, and Yongyi Mao
27. ICML 2025
28. 
    
29. Preserving Label Correlation for Multi-label Text Classification by Prototypical Regularizations
30. Fanshuang Kong, Richong Zhang, Xiaohui Guo, Junfan Chen, and Ziqiao Wang
31. WWW 2025
32. 
    
33. LH-Mix: Local Hierarchy Correlation Guided Mixup over Hierarchical Prompt Tuning
34. Fanshuang Kong, Richong Zhang and Ziqiao Wang
35. KDD 2025
36. 
    
37. Generalization Bounds via Conditional f-Information
38. Ziqiao Wang and Yongyi Mao
39. NeurIPS 2024
40. 
    
41. On f-Divergence Principled Domain Adaptation: An Improved Framework
42. Ziqiao Wang and Yongyi Mao
43. NeurIPS 2024
44. 
    
45. Two Facets of SDE Under an Information-Theoretic Lens: Generalization of SGD via Training Trajectories and via Terminal States
46. Ziqiao Wang and Yongyi Mao
47. UAI 2024
48. 
    
49. On Unsupervised Domain Adaptation: Pseudo Label Guided Mixup for Adversarial Prompt Tuning
50. Fanshuang Kong, Richong Zhang, Ziqiao Wang and Yongyi Mao
51. AAAI 2024
52. 
    
53. Cross-modal and Uni-modal Soft-label Alignment for Image-Text Retrieval
54. Hailang Huang, Zhijie Nie, Ziqiao Wang and Ziyu Shang
55. AAAI 2024
56. 
    
57. Sample-Conditioned Hypothesis Stability Sharpens Information-Theoretic Generalization Bounds
58. Ziqiao Wang and Yongyi Mao
59. NeurIPS 2023
60. 
    
61. Tighter Information-Theoretic Generalization Bounds from Supersamples
62. Ziqiao Wang and Yongyi Mao
63. ICML 2023 (Oral, top 2.2% of submissions)
64. 
    
65. Information-Theoretic Analysis of Unsupervised Domain Adaptation
66. Ziqiao Wang and Yongyi Mao
67. ICLR 2023
68. 
    
69. Over-Training with Mixup May Hurt Generalization
70. Zixuan Liu*, Ziqiao Wang* (equal contribution), Hongyu Guo, and Yongyi Mao
71. ICLR 2023
72. 
    
73. On the Generalization of Models Trained with SGD: Information-Theoretic Bounds and Implications
74. Ziqiao Wang and Yongyi Mao
75. ICLR 2022