Aleksander Madry

Aleksander Madry is a professor of computer science in MIT’s Department of Electrical Engineering and Computer Science and a principal investigator at the Computer Science and Artificial Intelligence Laboratory. He is also director of the MIT Center for Deployable Machine Learning. His research interests span algorithms, continuous optimization, the science of deep learning, and developing reliable, trustworthy and secure machine learning systems. Before coming to MIT, he was a postdoc at Microsoft Research New England and on the faculty of EPFL in Switzerland. His honors include an NSF Career Award, an Alfred P. Sloan Research Fellowship and the European Association for Theoretical Computer Science’s Presburger Award. Madry earned an undergraduate degree in theoretical physics and computer science from University of Wroclaw, and a PhD in computer science from MIT.

Publications

• Salman, H., Ilyas, A., Engstrom, L., Kapoor, A., Madry, A. (2020). Do Adversarially Robust ImageNet Models Transfer Better? Conference on Neural Information Processing Systems (NeurIPS).
• Engstrom, L., Ilyas, A., Santurkar, S., Tsipras, D., Janoos, F., Rudolph, L., Madry, A. (2019). Implementation Matters in Deep RL: A Case Study on PPO and TRPO. Proceedings of the International Conference on Learning Representations  (ICLR).
• Ilyas, A., Santurkar, S., Tsipras, D., Engstrom, L., Tran, B., Madry, A. (2019). Adversarial examples are not bugs, they are features. Advances in Neural Information Processing Systems, 125-136.
• Tsipras, D., Santurkar, S., Engstrom, L., Turner, A., Madry, A. (2019). Robustness may be at odds with accuracy. Proceedings of the International Conference on Learning Representations (ICLR).

Media

• March 7, 2020: MIT News, Doing Machine Learning the Right Way.
• July 31, 2019: MIT News, Why did my classifier just mistake a turtle for a rifle?
• December 14, 2018: MIT News, 3Q: Aleksander Madry on building trustworthy AI.