Seminar on Computational Learning and Adaptation

 Relational Temporal Difference Learning

Nima Asgharbeygi
Computational Learning Laboratory
Center for the Study of Language and Information
Stanford University

This talk shows how relational temporal difference learning is an effective approach to solving multi-agent Markov decision problems with large state spaces. An algorithm is presented that uses temporal difference reinforcement to learn a distributed value function represented over a conceptual hierarchy of relational predicates. Experimental results are described using two domains from the General Game Playing repository, in which we observe that our system achieves higher learning rates than non-relational methods. We also discuss related work and directions for future research.

This is joint work with David Stracuzzi and Pat Langley.

Return to the seminar schedule