AA1*: A Dynamic Incremental Network that Learns by Discrimination
An incremental learning algorithm for a special class of self-organising, dynamic networks is presented. Learning is effected by adapting both the function performed by the nodes and the overall network topology, so that the network grows (or shrinks) over time to fit the problem. Convergence is guaranteed on any arbitrary Boolean dataset and empirical generalisation results demonstrate promise.
KeywordsTraining Instance Negative Instance Node Selection Binary Decision Tree Node Table
Unable to display preview. Download preview PDF.
- C. Giraud-Carrier and T. Martinez. Using precepts to augment training set learning. In Proc. ANNES′93, 46–51. Google Scholar
- C. Giraud-Carrier and T. Martinez. Analysis of the convergence and generalization of AA1. JPDC, 1994 (to appear). Google Scholar
- L.O. Hall and S.G. Romaniuk. A hybrid connectionist, symbolic learning system. In Proc. AAAI′90, 783–788.Google Scholar
- T.R. Martinez . Adaptive Self-Organizing Networks. PhD thesis, University of California, Los Angeles, 1986.Google Scholar
- T.R. Martinez. Consistency and gener alization in incrementally trained connectionist networks. In Proc. International Symposium on Circuits and Systems, 706–709, 1990.Google Scholar
- T.R. Martinez, J.C. Barker, C. Giraud-Carrier. A generalizing adaptive discriminant network. In Proc. WCNN′93, I:613–616Google Scholar
- T.R. Martinez and D.M. Campbell. A self-adjusting dynamic logic module. JPDC, 11(4):303–313, 1991.Google Scholar
- T.R. Martinez and J.J. Vidal. Adaptive parallel logic networks. JPDC, 5(l):26–58, 1988.Google Scholar
- P.M. Murphy and D.W. Aha. UCI reposit ory of machine learning databases. University of California, Irvine, Department of Information and Computer Science, 1992.Google Scholar
- D. Ourston and R. Mooney. Changing the rules: A comp rehensive approach to theory refinement. In Proc. AAAI′90, 815–820.Google Scholar