Amartya Mitra

I am a final year grad student in physics at the University of California, Riverside and long-term intern at Mila since early 2019.

While the early part of my Ph.D. was spent on understanding quantum Hall systems, more recently my interests have shifted to understanding the theoretical foundations of deep learning. On that regard, my involvements at Mila are along the following three fronts:

a) Development of efficient optimization algorithms for min-max games – w/ Ioannis Mitliagkas

b) Training agents playing games with adversarial structure, to cooperate – w/ Aaron Courville

c) Understanding the generalization properties of over-parameterized NNs.- w/ Ioannis Mitliagkas