Accession Number : ADA183782

Title :   Multilayer Networks of Self-Interested Adaptive Units.

Descriptive Note : Final rept. Sep 83-Sep 86,

Corporate Author : MASSACHUSETTS UNIV AMHERST DEPT OF COMPUTER AND INFORMATION SCIENCE

Personal Author(s) : Barto,Andrew G

PDF Url : ADA183782

Report Date : Jul 1987

Pagination or Media Count : 149

Abstract : This report describes research directed toward refining and evaluating learning methods for multilayer networks of neuron-like adaptive units. We define a learning rule called the Associative Reward-Penalty, or A sub R-P, rule that has strong ties to both the theory of adaptive pattern classification and stochastic learning automata. We state a convergence result that has been proven for a single A sub R-P units can reliably learn nonlinear associative mappings. The behavior of these networks is discussed in terms of the collective behavior of stochastic learning automata in team decision problems. A number of methods for learning in multilayer networks are compared, including the A sub R-P method and the error back-propagation method. These methods, or variants of them, outperform the other methods applied to the test problem, with error back-propagation showing a significant speed advantage over the other methods. The A sub R-P and error back-propagation are compared and contrasted in terms of their respective approaches to gradient following.

Descriptors :   *ADAPTIVE SYSTEMS, *LEARNING, *COMPUTERIZED SIMULATION, *MATHEMATICAL ANALYSIS, ADAPTIVE SYSTEMS, AUTOMATA, BEHAVIOR, CLASSIFICATION, CONVERGENCE, DECISION MAKING, LAYERS, LEARNING, NETWORKS, PATTERNS, STOCHASTIC PROCESSES, TEAMS(PERSONNEL), VARIATIONS, ARTIFICIAL INTELLIGENCE, LEARNING CURVES, ALGORITHMS, ERRORS, NETWORK ANALYSIS(MANAGEMENT)

Subject Categories : Computer Systems
      Bionics

Distribution Statement : APPROVED FOR PUBLIC RELEASE