CONVERGENCE OF TEAMS AND HIERARCHIES OF LEARNING AUTOMATA IN CONNECTIONIST SYSTEMS

Learning algorithms for feedforward connectionist systems in a reinfor cement learning environment are developed and analyzed in this paper. The connectionist system is made of units of groups of learning automa ta, The learning algorithm used is the L(R-I) and the asymptotic behav ior of this algorithm is approximated by an Ordinary Differential Equa tion (ODE) for low values of the learning parameter, This is done usin g weak convergence techniques, The reinforcement learning model is use d to pose the goal of the system as a constrained optimization problem , It is shown that the ODE, and hence the algorithm exhibits local con vergence properties, converging to local solutions of the related opti mization problem, The three layer pattern recognition network is used as an example to show that the system does behave as predicted and rea sonable rates of convergence are obtained, Simulations also show that the algorithm is robust to noise.