Comparison of CMACs and radial basis functions for local function approximators in reinforcement learning

Author

Kretchmar, R. Matthew ; Anderson, Charles W.

Author_Institution

Dept. of Comput. Sci., Colorado State Univ., Fort Collins, CO, USA

Volume

2

fYear

1997

fDate

9-12 Jun 1997

Firstpage

834

Abstract

CMACs and radial basis functions are often used in reinforcement learning to learn value function approximations having local generalization properties. We examine the similarities and differences between CMACs, RBFs and normalized RBFs and compare the performance of Q-learning with each representation applied to the mountain car problem. We discuss ongoing research efforts to exploit the flexibility of adaptive units to better represent the local characteristics of the state space

Keywords

cerebellar model arithmetic computers; feedforward neural nets; function approximation; learning (artificial intelligence); CMAC; Q-learning; local function approximators; local generalization properties; mountain car problem; neural net; normalized RBF; radial basis functions; reinforcement learning; state space characteristics; value function approximations; Adaptive control; Art; Computer science; Function approximation; Learning; Programmable control; Shape; State-space methods; Table lookup; Tiles;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks,1997., International Conference on

Conference_Location

Houston, TX

Print_ISBN

0-7803-4122-8

Type

conf

DOI

10.1109/ICNN.1997.616132

Filename

616132