Tengyu Ma, B.S.

Tengyu Ma has a bachelor’s degree from Tsinghua University and is currently a second-year graduate student at Princeton University. Previously he won a silver medal at the International Mathematical Olympiad and was ranked 8th at MAA’s Putman Competition. Ma’s work seeks to develop efficient algorithms with provable guarantees for machine learning problems. One of his works gives polynomial time algorithms for a class of deep neural networks in the generative model and reveals interesting structures of neural networks with random weights. In another work, Ma and his co-authors showed that dictionary learning/sparse coding can be solved provably by minimizing a certain non-convex function using alternating minimization as well as simple neurally plausible algorithms.