An Ensemble Classification of Useful Melanoma Image Features

Abstract

In this study, we evaluate majority voting based ensemble classification combined with three and five common machine learning algorithms for enhancement on semi-automated system for detecting melanoma skin cancer. Melanoma is still increasingly becoming a deadly form of cancer around the world. In our previous study, the most useful six features from plain photographs of affected skin regions were identified after segmentation, feature extraction and selection, then used for classification and evaluated by using support vector machine with Gaussian radial basis kernel (SVM-RBF). However, the comparison with other machine learning algorithms may not be sufficient to claim its superiority. Thus in this paper, we use ensemble classification to examine further and show a slight improvement in the classification performance.