A neural network approach to adaptive state-space partitioning

An algorithm that learns to partition the state-space for a machine learned control application is presented, and the idea of competitive learning, a form of unsupervised learning, is introduced. A theoretical framework for a partitioning algorithm that is based on the neural network competitive learning model of T. kohonen's feature maps (1982, 1984) is developed. This algorithm is aimed at partitioning the BOXES machine learning algorithm. The goal was to enhance the functionality and the learning capability of BOXES by testing partitioning strategies. The modified BOXES algorithm did show an improved learning performance when compared to BOXES but needs to be tested against other known learning algorithms before its capabilities are judged.<<ETX>>

[1]  Teuvo Kohonen,et al.  Self-Organization and Associative Memory , 1988 .

[2]  Roman Bek,et al.  Discourse on one way in which a quantum-mechanics language on the classical logical base can be built up , 1978, Kybernetika.

[3]  GrossbergS..  Adaptive pattern classification and universal recoding , 1976 .

[4]  Katsuhisa Furuta,et al.  Control of unstable mechanical system Control of pendulum , 1976 .

[5]  Chen Hui-tang,et al.  Stabilization of a Double Inverted Pendulum by Analogue Controller , 1984 .

[6]  David Zipser,et al.  Feature Discovery by Competive Learning , 1986, Cogn. Sci..

[7]  Bernard Widrow,et al.  Adaptive switching circuits , 1988 .

[8]  Stephen Grossberg,et al.  The ART of adaptive pattern recognition by a self-organizing neural network , 1988, Computer.

[9]  Richard S. Sutton,et al.  Training and Tracking in Robotics , 1985, IJCAI.

[10]  Stephen Grossberg,et al.  Competitive Learning: From Interactive Activation to Adaptive Resonance , 1987, Cogn. Sci..

[11]  Charles W. Anderson,et al.  Strategy Learning with Multilayer Connectionist Representations , 1987 .

[12]  T. Kohonen.  Self-Organized Formation of Correct Feature Maps , 1982 .

[13]  Bing Zhang,et al.  Experiments in adaptive rule-based control , 1990, IEA/AIE '90.

[14]  Richard S. Sutton,et al.  Neuronlike adaptive elements that can solve difficult learning control problems , 1983, IEEE Transactions on Systems, Man, and Cybernetics.

[15]  S. C. Martin,et al.  AI Applied to Real Time Control: A Case Study , 1986 .

[16]  B. Zhang,et al.  A neural-net approach to supervised learning of pole balancing , 1989, Proceedings. IEEE International Symposium on Intelligent Control 1989.

[17]  Zhang Bing,et al.  The learned control of complex dynamic systems , 1991 .

[18]  Claude Sammut,et al.  Experimental Results from an Evaluation of Algorithms that Learn to Control Dynamic Systems , 1988, ML.

[19]  A. Grinnell,et al.  Introduction to Nervous Systems , 1978 .

[20]  Paul E. Utgoff,et al.  Learning to control a dynamic physical system , 1987, Comput. Intell..