Impact of soft clustering on classification of suspicious areas in digital mammograms
Peter McLeod, Brijesh Verma · 2008
This paper investigates a soft cluster based approach for determining the impact of soft clustering on the training of a neural network classifier for the classification of suspicious areas in digital mammograms. An approach is proposed that first creates soft clusters for each available class and then uses soft clusters to form subclasses within benign and malignant classes. The incorporation of soft clusters in the classification process is designed to increase the learning abilities and improve the accuracy of the classification system. The experiments using soft clusters based proposed approach and a standard neural network classifier have been conducted on a benchmark database. The results have been analysed and presented in this paper.