I am a Research Scientist in the Department of Computer Science at the University of Chicago working with Ben Zhao and Nick Feamster. My research focuses on robust and reliable machine learning. I am also interested in the application of machine learning to problems of societal interest. I completed my Ph.D. under the supervision of Prateek Mittal in the Department of Electrical and Computer Engineering at Princeton University. For an accessible overview of my research, see Research and for a full list of papers, see Publications.
- 09/2023: Paper on lower bounds for robust multi-class classifiers accepted to Neurips 2023 as a Spotlight!
- 07/2023: Two papers accepted to the ICML 2023 Workshop on Frontiers of Adversarial Machine Learning on lower bounds for multi-class models and feature extractors.
- 07/2023: The LEAF paper on drift mitigation in cellular networks is accepted to CoNEXT 2023. Congratulations Shinan!
- 05/2023: Paper on automated censorship detection is accepted to KDD 2023. Congratulations to the team!
- 10/2022: Submitted a comment to the FTC’s ANPR on Commercial Surveillance and Data Security with Emily Wenger.
- 09/2022: Two papers accepted to Neurips 2022 on analyzing learned representations from robustly trained networks and finding naturally backdoors in image datasets!
- 05/2022: Paper on poisoning attack forensics is accepted to USENIX 2022. Congratulations Shawn!
- 04/2022: Proposal on Fundamental Limits on the Robustness of Supervised Machine Learning Algorithms was awarded a grant from the C3.ai Digital Transformation Institute.
- 04/2022: Chapter on Adversarial Attacks for Anomaly Detection is forthcoming in the Springer Encyclopedia on Machine Learning and Data Science
- 01/2022: AISTATS 2022 paper on defending against model poisoning attack is up. Congratulations Ashwinee!
- 11/2021: Selected for the UChicago Rising Stars in Data Science.
- 09/2021: Paper on defenses against website fingerprinting attacks is accepted to AISec 2021. Congratulations Shawn!
- 07/2021: Paper on lower bounds on cross-entropy loss for classification with test-time attacks is appearing at IMCL 2021! Follow-up to our NeurIPS 2019 paper which introduced this line of work on fundamental lower bounds on loss in the presence of test-time attackers.
- 06/2021: Our paper on physical backdoor attacks is appearing at CVPR 2021.
- 05/2021: Runner-up for the Bede Liu Best Dissertation award from the Department of Electrical and Computer Engineering at Princeton University for my thesis The Role of Data Geometry in Adversarial Machine Learning.