Cloud-based Deep Learning Model for Classifying Skin Cancer

This study aims to proposes a CNN (Convolutional Neural Network) approach to the solve the problem of classifying skin cancer. Many cancer cases that are misdiagnosed in their early stages result in serious repercussions, including the patient’s death. In certain situations, individuals present with additional issues, which physicians diagnose as skin cancer. This results in the needless expenditure of time and funds on additional diagnostics. This paper discusses deep neural networks and transfer learning architectural difficulties and also focuses on optimizing our current solution which is to develop and deploy a CNN model using AWS, to outsource data-intensive computing. To train and test the network, we have used ISIC databases that are accessible to the public. The proposed model achieves ROC-AUC of 0.969, F1 score of 0.97, recall of 0.94, precision of 0.94, and an accuracy of 0.961, which is better than prior state-of-theart approaches.