Home
People
Publications
Research
Education
News & Events
Seminar Series
Contacts
Prospective Students
Seminars and News

GRASP Lab Seminar 2004-2005

December 3, 11:00 AM, Levine Hall 307.

Andrew Ng
Computer Science Department, Stanford

Reinforcement Learning and Apprenticeship Learning for Control.

Abstract: Many problems in control have unknown, stochastic, and highly non-linear dynamics, and offer significant challenges to classical control methods. Some of the key difficulties in these problems are that (i) It is often hard to write down, in closed form, a formal specification of the control task (for example, what is the objective function for "driving well"?), (ii) It is difficult to learn good control---as opposed to merely descriptive---models of the dynamics (cf. the "exploration problem" in reinforcement learning), and (iii) It is expensive to find closed-loop controllers for high dimensional, highly stochastic domains. In this talk, I will present formal results showing how (i) and (ii) can be efficiently addressed in the apprenticeship learning setting in which expert demonstrations of the task are available. I will also describe how efficient policy search algorithms can be applied to (iii). This talk will also draw from a number of case studies, including applications in autonomous helicopter flight, legged robot walking, snake robot locomotion, and monocular navigation. Joint work with Pieter Abbeel, Adam Coates, Wenmiao Lu, Yirong Shen and Nick Sivo.

Biography: Andrew Ng is an Assistant Professor in the Computer Science department at Stanford University. His research focuses on machine learning and statistical artificial intelligence, including their applications to robotic control and to text processing.

Seminar schedule

top of page