cwa0@gte-labs.CSNET ("Charles W. Anderson") (10/09/86)
The following is the abstract from my Ph.D. dissertation completed in August, 1986, at the University of Massachusetts, Amherst. Members of my committee are Andrew Barto, Michael Arbib, Paul Utgoff, and William Kilmer. I welcome all comments and questions. Chuck Anderson GTE Laboratories Inc. 40 Sylvan Road Waltham, MA 02254 617-466-4157 cwa0@gte-labs Learning and Problem Solving with Multilayer Connectionist Systems The difficulties of learning in multilayered networks of computational units has limited the use of connectionist systems in complex domains. This dissertation elucidates the issues of learning in a network's hidden units, and reviews methods for addressing these issues that have been developed through the years. Issues of learning in hidden units are shown to be analogous to learning issues for multilayer systems employing symbolic representations. Comparisons of a number of algorithms for learning in hidden units are made by applying them in a consistent manner to several tasks. Recently developed algorithms, including Rumelhart, et al.'s, error back-propagation algorithm and Barto, et al.'s, reinforcement-learning algorithms, learn the solutions to the tasks much more successfully than methods of the past. A novel algorithm is examined that combines aspects of reinforcement learning and a data-directed search for useful weights, and is shown to out perform reinforcement-learning algorithms. A connectionist framework for the learning of strategies is described which combines the error back-propagation algorithm for learning in hidden units with Sutton's AHC algorithm to learn evaluation functions and with a reinforcement-learning algorithm to learn search heuristics. The generality of this hybrid system is demonstrated through successful applications to a numerical, pole-balancing task and to the Tower of Hanoi puzzle. Features developed by the hidden units in solving these tasks are analyzed. Comparisons with other approaches to each task are made.