CS 294-7, Spring 2003
Instructor Stuart Russell
727 Soda Hall,
(510) 642 4964
Office hours Monday 10.00-12.00.
Lecture: Wed 12.30-2.30
Location: 405 Soda
Suggested prerequisites: CS188 or equivalent, or permission of
This class will look at reinforcement learning, i.e.,
learning how to behave given experience in an environment
that provides feedback in the form of rewards and penalties.
Topics will include
Discussion will focus on readings from the following:
There will be two or three homework assignments and a substantial
project or analytical paper.
- Foundations: Markov decision processes and dynamic programming
- Model-based reinforcement learning
- Model-free reinforcement learning: Q-learning
- Function approximation
- Policy search methods
- Hierarchical reinforcement learning
- Applications of reinforcement learning