Title :
A genetic algorithm for reinforcement learning
Author :
Zhao, Long ; Liu, Zemin
Author_Institution :
Beijing Univ. of Posts & Telecommun., China
Abstract :
A main difficulty for applying reinforcement learning methods in practice is the large action space problem. In this paper, by means of the concept of genetic optimization, we have proposed a new reinforcement learning algorithm. Our genetic algorithm for reinforcement learning has solved the large action space problem quite well
Keywords :
genetic algorithms; learning (artificial intelligence); neural nets; genetic algorithm; genetic optimization; large action space problem; reinforcement learning; Adaptive control; Computational complexity; Equations; Genetic algorithms; Learning automata; Neural networks; Neurons; Programmable control; State-space methods; Stochastic processes;
Conference_Titel :
Neural Networks, 1996., IEEE International Conference on
Conference_Location :
Washington, DC
Print_ISBN :
0-7803-3210-5
DOI :
10.1109/ICNN.1996.549044
