Committee: 

Dr. Theodorou, Advisor

Dr. Bloch, Co-Advisor   

Dr. Romberg, Chair

Dr. Tao

Abstract: The objective of the proposed research is to bridge the connection between deep learning and stochastic optimal control. Precisely,  this proposal is trying to leverage the systematic stochastic optimal control theory to shed light on deep learning models, and oppositely, taking the advantage of the universal approximation property of deep learning to numerically solve intractable problems in large scale stochastic optimal control problems.