Assessment of fuzzy min-max neural networks for classification tasks

Abstract

Statistical methods have been used in order to classify data from random samples. Generally, if we know the statistical distribution of the data, we can utilize specific classifiers designed for that distribution and anticipate good results. This work assesses the accuracy of Fuzzy Min-Max Neural Network (FMM) and Enhanced Fuzzy Min-Max Neural Network (EFMM) classifiers in classification tasks using data from five different statistical distributions: Negative Binomial, Logistic, Log-Normal, Gamma, and Weibull. Results of the assessment are provided and show different accuracies based on the statistical distribution of the data. This study presents a novel approach to the classification of statistical distributions by presenting two classifiers, namely FMM and EFMM Neural Networks, capable of classifying the above statistical distributions.