Hi! I am Wenxiao Wang (汪文潇), a Computer Science Ph.D. student at University of Maryland, working with Prof. Soheil Feizi. I received my B.S. degree in Computer Science from Yao Class, Tsinghua University in 2020.
I was a research intern at Sony AI (summer 2023), working with Dr. Weiming Zhuang and Dr. Lingjuan Lyu in Privacy-Preserving Machine Learning (PPML) team; I was a research intern at Bytedance (summer 2022), working with Dr. Linjie Yang, Dr. Heng Wang and Dr. Yu Tian; a research assistant at IIIS, Tsinghua University (2020-2021), working with Prof. Hang Zhao in his MARS Lab; a visiting student researcher at UC Berkeley (2019), working with Dr. Xinyun Chen, Prof. Ruoxi Jia and Prof. Dawn Song; an intern in Bytedance AI Lab (2018), working with Dr. Yi He and Prof. Lei Li.
Preprints
Towards Fundamentally Scalable Model Selection: Asymptotically Fast Update and Selection
Wenxiao Wang, Weiming Zhuang and Lingjuan Lyu
[arxiv]
On Practical Aspects of Aggregation Defenses against Data Poisoning Attacks
Wenxiao Wang and Soheil Feizi
[arxiv]
Can AI-Generated Text be Reliably Detected?
Vinu Sankar Sadasivan, Aounon Kumar, Sriram Balasubramanian, Wenxiao Wang and Soheil Feizi
Media Coverage:
[Washington Post]
[Wired]
[New Scientist]
[The Register]
[TechSpot]
[UMD Science]
[arxiv]
Publications
Robustness of AI-Image Detectors: Fundamental Limits and Practical Attacks
Mehrdad Saberi, Vinu Sankar Sadasivan, Keivan Rezaei, Aounon Kumar, Atoosa Chegini, Wenxiao Wang and Soheil Feizi
Media Coverage:
[Wired]
[MIT Tech Review]
[Bloomberg News]
[The Register]
International Conference on Learning Representations (ICLR), 2024.
[paper]
DRSM: De-Randomized Smoothing on Malware Classifier Providing Certified Robustness
Shoumik Saha, Wenxiao Wang, Yigitcan Kaya, Soheil Feizi and Tudor Dumitras
International Conference on Learning Representations (ICLR), 2024.
[paper]
Temporal Robustness against Data Poisoning
Wenxiao Wang and Soheil Feizi
Conference on Neural Information Processing Systems (NeurIPS), 2023.
[paper]
Spuriosity Rankings: Sorting Data for Spurious Correlation Robustness
Mazda Moayeri, Wenxiao Wang, Sahil Singla and Soheil Feizi
Conference on Neural Information Processing Systems (NeurIPS), 2023. [spotlight]
[paper]
Lethal Dose Conjecture on Data Poisoning
Wenxiao Wang, Alexander Levine and Soheil Feizi
Conference on Neural Information Processing Systems (NeurIPS), 2022.
[paper] [code]
Improved Certified Defenses against Data Poisoning with (Deterministic) Finite Aggregation
Wenxiao Wang, Alexander Levine and Soheil Feizi
International Conference on Machine Learning (ICML), 2022.
[paper] [code]
On Feature Decorrelation in Self-Supervised Learning
Tianyu Hua*, Wenxiao Wang*, Zihui Xue, Sucheng Ren, Yue Wang, Hang Zhao (*equal contribution)
International Conference on Computer Vision (ICCV), 2021. [oral]
[paper] [code]
DPlis: Boosting Utility of Differentially Private Deep Learning via Randomized Smoothing
Wenxiao Wang, Tianhao Wang, Lun Wang, Nanqing Luo, Pan Zhou, Dawn Song, Ruoxi Jia
Privacy Enhancing Technologies Symposium (PETS), 2021.
[paper] [code]
REFIT: A Unified Watermark Removal Framework For Deep Learning Systems With Limited Data
Xinyun Chen*, Wenxiao Wang*, Yiming Ding, Chris Bender, Ruoxi Jia, Bo Li, Dawn Song (*equal contribution)
ACM Asia Conference on Computer and Communications Security (AsiaCCS), 2021.
[paper] [code]
The Secret Revealer: Generative Model Inversion Attacks Against Deep Neural Networks
Yuheng Zhang*, Ruoxi Jia*, Hengzhi Pei, Wenxiao Wang, Bo Li, Dawn Song (*equal contribution)
Conference on Computer Vision and Pattern Recognition (CVPR), 2020. [oral]
[paper] [code]
Leveraging Unlabeled Data for Watermark Removal of Deep Neural Networks
Xinyun Chen*, Wenxiao Wang*, Yiming Ding, Chris Bender, Ruoxi Jia, Bo Li, Dawn Song (*equal contribution)
ICML Workshop on Security and Privacy of Machine Learning, 2019.
[paper]
Talks
- Temporal Robustness against Data Poisoning, AI TIME Youth PhD Talk, November 2023.
- Lethal Dose Conjecture: From Few-shot Learning to Potentially Nearly Optimal Defenses against Data Poisoning, TMLR Group, Hong Kong Baptist University, December 2022.
- Lethal Dose Conjecture on Data Poisoning, AI TIME Youth PhD Talk, November 2022.
- Improved Certified Defenses against Data Poisoning with (Deterministic) Finite Aggregation, AI TIME Youth PhD Talk, August 2022.
Services
Program Committee / Reviewer of:
- TPAMI
- NeurIPS 2022, 2023
- ICML 2022, 2023 (Outstanding Reviewer in 2022)
- ICLR 2023
- ICCV 2023
- Workshop on Adversarial Robustness In the Real World (ECCV 2022,ICCV 2021)
- Workshop on Socially Responsible Machine Learning (ICML 2021)
- Workshop on Adversarial Machine Learning in Real-World Computer Vision Systems and Online Challenges (CVPR 2021)
- Workshop on Security and Safety in Machine Learning Systems (ICLR 2021)