Perbandingan Kinerja Model Convolutional Neural Network pada Klasifikasi Kanker Kulit

Abstract

Skin cancer is one of the diseases that greatly affects the quality of human life and can be potentially bad if not detected and treated quickly. In an effort to improve the early detection and accuracy of skin cancer diagnosis, this study aims to design a skin cancer classification system using deep learning. The research method involves the use of deep learning-based artificial neural networks, especially convolutional neural networks (CNN), which have been proven effective in image processing. The images in the dataset include various types of skin cancer, including melanoma and basal cell carcinoma. In this study, a skin cancer classification system was built using The-HAM10000 Dataverse, utilizing the CNN algorithm and the performance of the system was modeled using InceptionV3, and EfficientNetB0. The results of these two methods on the dataset show that InceptionV3 has an accuracy of 0.7681, and EfficientNetB0 of 0.9809, the results of this accuracy make the EfficientNetB0 system better for skin cancer detection, so this finding makes that the accuracy at this level will classify as much as 327 data only, therefore the data trained and tested will produce an accuracy and error from its validation which makes EfficientNetB0 an option for classification based on the CNN method.