A Caltech Library Service

New Frameworks for Structured Policy Learning


Le, Hoang Minh (2020) New Frameworks for Structured Policy Learning. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/rz4w-k233.


Sequential decision making applications are playing an increasingly important role in everyday life. Research interest in machine learning approaches to sequential decision making has surged thanks to recent empirical successes of reinforcement learning and imitation learning techniques, partly fueled by recent advances in deep learning-based function approximation. However in many real-world sequential decision making applications, relying purely on black box policy learning is often insufficient, due to practical requirements of data efficiency, interpretability, safety guarantees, etc. These challenges collectively make it difficult for many existing policy learning methods to find success in realistic applications.

In this dissertation, we present recent advances in structured policy learning, which are new machine learning frameworks that integrate policy learning with principled notions of domain knowledge, which spans value-based, policy-based, and model-based structures. Our framework takes flexible reduction-style approaches that can integrate structure with reinforcement learning, imitation learning and robust control techniques. In addition to methodological advances, we demonstrate several successful applications of the new policy learning frameworks.

Item Type:Thesis (Dissertation (Ph.D.))
Subject Keywords:Structured Policy Learning, Policy Learning, Reinforcement Learning, Imitation Learning, Safe Machine Learning
Degree Grantor:California Institute of Technology
Division:Engineering and Applied Science
Major Option:Computing and Mathematical Sciences
Awards:Amori Doctoral Prize in CMS, 2020. MIT Sloan Conference 2017, Best paper runner up. Amazon Graduate Fellow, 2017.
Thesis Availability:Public (worldwide access)
Research Advisor(s):
  • Yue, Yisong
Thesis Committee:
  • Wierman, Adam C. (chair)
  • Anandkumar, Anima
  • Daumé, Hal, III
  • Yue, Yisong
Defense Date:22 October 2019
Funding AgencyGrant Number
Intel CorporationUNSPECIFIED
Northrop Grumman CorporationUNSPECIFIED
Record Number:CaltechTHESIS:06092020-121556493
Persistent URL:
Related URLs:
URLURL TypeDescription research website Article adapted for Ch. 3 adapted for Ch. 4 Article adapted for Ch. 5 Article adapted for Ch. 6 Article adapted for Ch. 7 Article adapted for Ch. 8
Le, Hoang Minh0000-0002-5521-5856
Default Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:13809
Deposited By: Hoang Le
Deposited On:11 Jun 2020 22:12
Last Modified:10 Dec 2020 00:04

Thesis Files

PDF - Final Version
See Usage Policy.


Repository Staff Only: item control page