Extended Self Organizing Map with Probabilistic Neural Network for Pattern Classification Problems

摘要

This paper presents a hybrid classifier based on extended Self Organizing Map with Probabilistic Neural Network. In this approach, at first we use feature extraction technique of Self Organizing Map to achieve topological ordering in the input data pattern. Then, with the use of Gaussian function, we obtain a better representation of the input dataset. After that, Probabilistic Neural Network is used to classify the input data. We have tested the proposed scheme on Iris, Glass, Breast Cancer Wisconsin, Wine, Ionosphere, Liver (BUPA), Sonar, Thyroid, and Vehicle data sets. The experimental results show better recognition accuracy of the proposed model than that of traditional Probabilistic Neural Network based classifier.