Home // ICAS 2018, The Fourteenth International Conference on Autonomic and Autonomous Systems // View article

A Study of the of Evolutionary Strategies Impact on Performance of Reinforcement Learning Autonomous Agents

Authors:
Ventseslav Shopov
Vanya Markova

Keywords: Autonomous Agents; Deep Reinforcement Learning; Evolutionary Computation.

Abstract:
Abstract—Algorithms for evolutionary computation, are applied in reinforcement learning autonomous agent to discover high-performing reinforcement-learning policies. Evolutionary reinforcement-learning approaches allow the agent to find good representations and cope with partial environment observability. We have compared the performance of classic reinforcement learning and evolutionary augmented autonomous agent in area of sequential games.

Pages: 48 to 51

Copyright: Copyright (c) IARIA, 2018

Publication date: May 20, 2018

Published in: conference

ISSN: 2308-3913

ISBN: 978-1-61208-634-7

Location: Nice, France

Dates: from May 20, 2018 to May 24, 2018