Real-Life Reinforcement Learning

Real-life reinforcement learning is the topic of building computer systems that learn to improve a measure of performance on a task by interacting with the real world. It was the topic of a 2004 AAAI Fall Symposium in Washington DC and describes work being carried out all over the world.

This web page is intended to be a hub pointing to various kinds of activity on the topic of RL RL.

Cool Links

RL Myths: At the symposium, Satinder Singh and John Langford had the idea of debunking some harmful myths about reinforcement learning.
Real-life Reinforcement Learning: The Reality Series (14M): This movie was shown at the plenary session of the symposium. It (somewhat) documents the first day of activities.

Submitted Talks

These are materials from some talks given at the symposium.

Introduction to Real-life Reinforcement learning, Michael Littman. Includes the symposium schedule.
Real-Life Multiagent Reinforcement, Michael Littman.

Submitted Papers

These are the papers submitted to the symposium working notes. They are unedited, so please do not cite!

coming soon

Lab Links

Autonomous Learning Laboratory: University of Massachusetts
Reinforcement Learning and AI: University of Alberta
Reinforcement Learning at the University of Michigan: University of Michigan
Rutgers Laboratory for Real-Life Reinforcement Learning (RL³): Rutgers University

Contact Michael Littman (mlittman@cs.rutgers.edu) to contribute material. Thanks!