Nuevas propuestas en el ámbito de los algoritmos genéticos distribuidos para la extracción de reglas de clasificación

Abstract

This research it is based on the combination of distributed genetic models with the genetic algorithms for rule learning in order to produce a distributed learning model flexible and scalable in the scenery of imbalanced datasets. A new algorithm has been developed named EDGAR (Efficient Distributed Genetic Algorithm for Rule Extraction). This algorithm produces high quality rule classifiers in accuracy and size. EDGAR combines data distribution with genetic distributed models in order to solve high dimensional data problems and imbalanced data problems with competitive results in the context of supervised genetic algorithm for classification . New techniques has been developed allowing this proposal to handle directly