CS 672: Learning and Sequential Decision Making

Rutgers University
Spring 2004
Michael L. Littman

Time: Monday, Wednesday 4:30-5:50
Place: Rutgers, Hill Center 482/484
Semester: Spring 2004

Michael's office hours: Hill 409, Wednesday 2pm-3pm and by appointment (mlittman@cs.rutgers.edu).

Description: Through a combination of classic papers and more recent work, the course will explore automated decision making from a computer-science perspective. It will examine efficient algorithms, where they exist, for single agent and multiagent planning as well as approaches to learning near-optimal decisions from experience. Topics will include Markov decision processes, stochastic and repeated games, partially observable Markov decision processes, and reinforcement learning. Of particular interest will be issues of generalization, exploration, and representation. Each student will be expected to present a published research paper and will participate in a group project to create a reinforcement-learning agent to compete in a video game environment. Participants should have taken a graduate-level computer science course and should have some exposure to reinforcement learning from a previous computer-science class or seminar; check with instructor if not sure. This is the first time the course is being offered.

News (most recent first)

4/26/04: RARS example tracks due Wednesday.
4/20/04: RARS group presentations scheduled for May 2nd. Final is scheduled for May 12th.
3/29/04: No class 4/5/04, meet in RARS groups. Guest lecture 4/12 (Russ Greiner).
3/01/04: We're teaming up with the algorithms class to cover Littman and Stone (2004) and Hart and Mas-Colell (2000). We probably won't have time for the second paper, though.
2/25/04: Here's a pointer to that RARS report from Berkeley. Oh, here's another one.
2/9/04: We're talking about video games today.
2/8/04: We're now meeting in Hill.
1/21/04: First class meets today.

Topics and Papers

Throughout the semester we will be reading sections of the RL survey by Kaelbling, Littman, Moore (1996).

Markov decision processes and algorithms. Survey, Sections 1 and 3. Littman, Dean, Kaelbling (1995).
TD-lambda. Survey, Section 4.1. Sutton (1988).
Q-learning/Convergence. Survey, remainder of Section 4, Section 5. Littman and Szepesvári (1996).
Eligibility traces. Singh and Sutton (1996).
Shaping. Ng, Harada, Russell (1999).
Exploration. Survey, Section 2. Fong (1995). Kearns and Singh (1998).
Repeated Games. Littman and Stone (2004). Hart and Mas-Colell (2000). Greenwald and Jafari (2003).
Generalization and convergence. Survey, Sections 6.1, 6.2. Baird (1995). Gordon (1995).
Partially observable environments. Survey, Section 7. Cassandra, Kaelbling, Littman (1994). Cassandra, Littman, Zhang (1997).
RL in POMDPs. Chrisman (1992). Loch and Singh (1998).
Hierarchy. Survey, remainder of Section 6. Dietterich (1998).
Policy search. Baxter and Bartlett (1999) .
Non-stationary environments. Sutton (1990).
Instance-based RL. Ormoneit and Sen (1999).
Applications. Survey, Sections 8 and 9. Crites and Barto (1996), Tesauro (1992).

RL Links

The URL for this page is http://www.cs.rutgers.edu/~mlittman/courses/rl04/.