Aplicando redes de aprendizado profundo e algoritmos de seleção dinâmica para classificar imagens de câncer de pele

Abstract

This work aims to leverage deep learning network-based methods for extracting attributes from skin cancer images from the HAM10000 dataset. By employing deep learning architectures, the study seeks to identify and highlight the main characteristics of images, to use them in the classifier. The ultimate goal involves developing a dynamic selection framework to classify these images based on different types of skin cancer. This approach surpasses traditional ones, being ahead of the use of deep learning networks and machine and ensemble learning methods for classification.