An iterative fuzzy prototype induction algorithm

摘要: Lazy learning methods have been proved useful when dealing with problems in which the examples multiple local functions. These are related selection, for training purposes, of a subset examples, ancl making some linear combination to generate output. On other hand, neural network eager that high nonlinear behavior. In this work, lazy method is proposed Radial Basis Neural Networks order improve both, generalization capability those networks specific domains, and performance classical mnethods. A comparison mnethods, RBNN trained as usual made, new approach shows good results two test real life problem an artificial domain.