eTS with Rule Base Simplification and Reduction for Knowledge Extraction from Data Sets

Authors

Abstract

Knowledge extraction from data sets using fuzzy logic is only effective if interpretability is one of the properties of the obtained fuzzy rules. When fuzzy models are developed from data (data-driven approach) it is difficult to avoid some degree of redundancy and unnecessary complexity in the obtained membership functions, preventing to give some semantic meaning to the results. On-line mechanisms for merging membership functions and rule base simplification are studied in order to improve the interpretability of the fuzzy models. This allows the minimization of redundancy and complexity of the models during their development, contributing automatically to the transparency of the obtained rules. The on-line learning technique used is the evolving Takagi-Sugeno (eTS) fuzzy models, based on an on-line learning algorithm that recursively develops the model structure and parameters. Illustrative results for a benchmark data set, the Box-Jenkins time-series prediction, are presented.

Subject

multidimensional sclaing; refinery

Related Project

CLASSE - Classificação Sintética para Supervisão Industrial (Synthetic Classification for Industrial Supervision)

Conference

CLAIO- Congresso Latino Americano de Investigacion de Operationes, September 2008