A Comparative Study of Ensemble Deep Learning Models for Skin Cancer Detection

This paper presents a comparison study between two ensemble learning models for the early detection of skin cancer. Our research is based on The HAM10000 dataset, a diverse collection of multi-sourced images of common pigmented skin lesions which consists of 10015 dermatoscopic images. The main focus of this study is to compare the performance of a CNN ensemble model and a SVM ensemble model in terms of accuracy, sensitivity, specificity, and computation complexity. Our results showed that the ensembled CNN and SVM models performed better than individual models in terms of accuracy. The main contribution of this paper is the production of a comparative study of different ensemble models that are designed to reduce prediction error using decomposition and reconstruction techniques for skin cancer detection.