Publications

BridgePure: Limited Protection Leakage Can Break Black-Box Data Protection. Conference
Yihan Wang*, Yiwei Lu*, Xiao-Shan Gao, Gautam Kamath, Yaoliang Yu.
Proceedings of the 39th Conference on Neural Information Processing Systems (NeurIPS 2025).
Not All Samples Are Equal: Quantifying Instance-level Difficulty in Targeted Data Poisoning. Preprint

William Xu, Yiwei Lu, Yihan Wang , Matthew Y.R. Yang, Zuoqiu Liu, Gautam Kamath, Yaoliang Yu.
MUC: Machine Unlearning for Contrastive Learning with Black-box Evaluation. Journal

Yihan Wang*, Yiwei Lu*, Guojun Zhang, Franziska Boenisch, Adam Dziedzic, Yaoliang Yu, Xiao-Shan Gao.

Transactions on Machine Learning Research (TMLR), 2025
Provable Robust Overfitting Mitigation in Wasserstein Distributionally Robust Optimization. Conference
Shuang Liu, Yihan Wang, Xiao-Shan Gao.
Proceedings of the 13th International Conference on Learning Representations (ICLR 2025)
Data-Dependent Stability Analysis of Adversarial Training. Journal
Yihan Wang, Shuang Liu, Xiao-Shan Gao.
Neural Networks, Volume 183, 2025
Efficient Availability Attacks against Supervised and Contrastive Learning Simultaneously. Conference
Yihan Wang, Yifan Zhu, Xiao-Shan Gao.
Proceedings of the 38th Conference on Neural Information Processing Systems (NeurIPS 2024).
On the Robustness of Neural Networks Quantization against Data Poisoning Attacks. Workshop

Yiwei Lu, Yihan Wang, Guojun Zhang, Yaoliang Yu.

ICML 2024 Next Generation of AI Safety Workshop.
Game-Theoretic Unlearnable Example Generator. Conference
Shuang Liu, Yihan Wang, Xiao-Shan Gao.
Proceedings of the 38th Annual AAAI Conference on Artificial Intelligence (AAAI 2024).
Adversarial Parameter Attack on Deep Neural Networks. Conference

Lijia Yu, Yihan Wang, Xiao-Shan Gao.

Proceedings of the 40th International Conference on Machine Learning (ICML 2023).
Restore Translation Using Equivariant Neural Networks. Conference
Yihan Wang, Lijia Yu, Xiao-Shan Gao.
Proceedings of the 30th International Conference on Neural Information Processing (ICONIP 2023).