I study adversarial machine learning from a learning theory perspective. My supervisors James Worrell, Varun Kanade and Marta Kwiatkowska and I recently published our work on the sample and computational complexity of robust classification with a follow-up on the special case of non-monotone decision lists. I am particularly interested in identifying thresholds for robust learning, namely when can efficient robust learning be guaranteed.

Thesis


Journal Publications

Conference Publications

For the most up-to-date list, please see Google scholar.