Intelligent optimization techniques for fuzzy logic based data Classification

Abstract

Classification is one of the important characteristics of Data Mining newlineconcepts that aims at building a classifier model for describing the important newlinedata classes and provides a better understanding of data at large Mostly newlineEngineering domains which includes Pattern Recognition Image Processing newlineGeo statistics Computer Networks Fault Detection Medical Diagnosis Bioinformatics newlineand so on involves data classification newlineGenetic Algorithm GA is proposed as a tool for the optimal design of newlinefuzzy classifier model The conventional Binary coded GA has Hamming newlinecliff problem which sometimes cause difficulties when coding continuous newlinevariables Also for discrete variables with total number of permissible newlinechoices not equal to 2k it becomes difficult to use a fixed length binary newlinecoding to represent all permissible values To address these issues Realcoded newlineGA RGA using Tournament selection BLX and#945; crossover and Nonuniform newlinemutation is proposed to design the fuzzy classifier newline newline