Stochastic Optimization Using Continuous Action-Set Learning Automata

In this paper, an adaptive random search method, based on continuous action-set learning automata, is studied for solving stochastic optimization problems in which only the noise-corrupted value of a function at any chosen point in the parameter space is available. First, a new continuous action-set learning automaton is introduced and its convergence properties are studied. Then, applications of this new continuous action-set learning automata to the minimization of a penalized Shubert function and pattern classification are presented