My research is broadly concerned with the theoretical foundations of machine learning and optimization; I am especially interested in the statistical and computational limits and tradeoffs of learning. Specific topics I like to think about include online learning and stochastic optimization, statistical learning theory, non-convex optimization, deep learning, and algorithm design, as well as concentration inequalities and game theory.
Inference on Graphs with Noisy Side Information:
Tight Rates and Efficient Algorithms
Dylan J. Foster, Daniel Reichman, and Karthik Sridharan.
Dylan J. Foster, Zhiyuan Li, Thodoris Lykouris, Karthik Sridharan, and Éva Tardos.
NIPS 2016. Preliminary version at Ad Auctions workshop at EC 2016.
Dylan J. Foster, Alexander Rakhlin, and Karthik Sridharan.
NIPS 2015 (with spotlight presentation).
Theory Seminar, Cornell University, October 2016
Machine Learning Seminar, New York University, October 2016
Algorithms Seminar, Google Research NYC, August 2016
AI Seminar, Cornell University, October 2016
Machine Learning Seminar, Google Research NYC, June 2016
Invited Talk, Learning from Easy Data II, NIPS Workshop, December 2015
Spotlight Presentation, NIPS 2015, December 2015
I can be reached at djf244 at cornell dot edu. My office is in the Theory Lab at Gates 336.
Copyright © 2016.