A Collective Model of Learning Automata with N-Cooperative Random Environments

Abstract

In this paper, we consider a collective model of learning automata operating on N-static random environments. Each automaton has a behavioural tactic directed towards the realization of its own goal, estimating the magnitude of the expected value of utility function that depends explicity on the automaton's strategy and the corresponding responses from environments. From game theoretic viewpoint, we can construct several types of environments for this model.
In here, a non-cooperative game on N-cooperative environments is proposed. For this model, the definitions, evaluation metrics and a distance diminishing reinforcement scheme are introduced. As an application, a simple numerical example is shown.