1995 Volume 115 Issue 8 Pages 953-960
There exist many types of cooperativity in the mathematical models and computer simulations of neural networks. In a previous paper we presented a reinforcement learning network with a different type of cooperativity from that of neural networks, using self-interested learning automata. In this paper we generalize our previous reinforcement learning algorithm in order to solve some learning problems with variable random environments, and also present a new algorithm based on Bayes decision theory. Simulation results show how cooperative activity in reinforcement learning networks can solve learning control problems.
The transactions of the Institute of Electrical Engineers of Japan.C
The Journal of the Institute of Electrical Engineers of Japan