On the learning behaviors of variable-structure stochastic automaton in the general n-teacher environment
Norio Baba · IEEE Transactions on Systems Man and Cybernetics · 1983
Learning behaviors of a stochastic automaton operating in a multiteacher environment are considered. As a generalized form of theLR-Ischeme, theGLR-Ischeme is proposed as a reinforcement scheme in a multiteacher environment. It is shown that theGLR-Ischeme is absolutely expedient and ϵ-optimal in the generaln-teacher environment. Learning behaviors of theGLR-Ischeme are simulated by computer and the results indicate the effectiveness of theGLR-Ischeme.