Multimodal searching technique based on learning automata with continuous input and changing number of actions

This paper describes a multimodal searching technique based on a stochastic automaton. The environment where the automaton operates corresponds to the function to be optimized which is assumed to be unknown function of a single parameter x. The admissible region of x is quantized into N subsets. The environment response is continuous (S-model). The complete set of actions of the automaton is divided into nonempty subsets. The action set is changing from instant to instant and is selected based on a probability distribution. These actions are in turn associated with the discrete values of the parameter x. Convergence and convergence rate results are presented. Simulation results illustrate the performance of this searching technique.