Atul Prakash

By |

My research interests include security, privacy, and adversarial machine learning.
More information about some of the current projects can be found at

9.9.2020 MIDAS Faculty Research Pitch Video.

An adversarial testing pipeline to fool machine learning classifiers. A STOP sign is being modified using stickers so that a state-of-the-art classifier is fooled into thinking it is a 45 SPEED LIMIT sign. For more details, visit (Robust Physical Perturbations tab)