Deep Convolutional Neural Networks for Skin Lesions Classification via Ensemble Strategy

Abstract

The early diagnosis of skin cancer is an urgent problem in public health, which helps to improve the survival rate of patients. In order to improve the model generalization ability, we propose an ensemble model, which first uses multiple different neural network models to train each model on the training set and then integrates the output of different models to form a rich feature representation to classify skin lesions. The experimental results show that the proposed ensemble model is effective, and its performance is better than the single model, with an accuracy of 89.5%.