The Performance of Deep and Conventional Machine Learning Techniques for Skin Lesion Classification

Abstract

Skin lesion is any abnormalities occurring to the skin's tissue in terms of size, texture, shape, and color. It can be a sign of autoimmune disorders, diabetes, etc. It can be a potentially huge threat to human health leading to skin cancer if not diagnosed early enough and treated. In this paper, we studied how machine learning algorithms can help detect Skin Lesion based on the images in Skin Legion dataset. Our study highlights the effectiveness of deep learning algorithms by utilizing the state-of-the-art CNN models which performed better in terms of classification performance than ML traditional methods comparatively.