Adaptive Bases for Reinforcement Learning
Machine Learning
2010-05-04 v1 Artificial Intelligence
Abstract
We consider the problem of reinforcement learning using function approximation, where the approximating basis can change dynamically while interacting with the environment. A motivation for such an approach is maximizing the value function fitness to the problem faced. Three errors are considered: approximation square error, Bellman residual, and projected Bellman residual. Algorithms under the actor-critic framework are presented, and shown to converge. The advantage of such an adaptive basis is demonstrated in simulations.
Cite
@article{arxiv.1005.0125,
title = {Adaptive Bases for Reinforcement Learning},
author = {Dotan Di Castro and Shie Mannor},
journal= {arXiv preprint arXiv:1005.0125},
year = {2010}
}