A Self-Made Agent Based on Action-Selection - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2003

A Self-Made Agent Based on Action-Selection

Olivier Buffet
Alain Dutech

Abstract

Some agents have to face multiple objectives simultaneously. In such cases, and considering partially observable environments, classical Reinforcement Learning (RL) is prone to fall in pretty low local optima, only learning straightforward behaviors. We present here a method that tries to identify and learn independent ``basic'' behaviors solving separate tasks the agent has to face. Using a combination of these behaviors (an action-selection algorithm), the agent is then able to efficiently deal with various complex goals in complex environments.
No file

Dates and versions

inria-00099828 , version 1 (26-09-2006)

Identifiers

  • HAL Id : inria-00099828 , version 1

Cite

Olivier Buffet, Alain Dutech. A Self-Made Agent Based on Action-Selection. Sixth European Workshop on Reinforcement Learning - EWRL-6 2003, 2003, Nancy, France, pp.47-48. ⟨inria-00099828⟩
77 View
0 Download

Share

Gmail Facebook Twitter LinkedIn More