Detection and optimization of skin cancer using deep learning

Abstract Convolutional Neural Network (CNN) is a branch of deep learning which has been one popular methods in different applications, especially medical field. In this study, an optimized CNN model built using the random search optimization to classify seven types skin cancer, namely, basal cell carcinoma, melanoma, dermatofibroma, vascular lesion, melanocytic nevus, actinic keratosis and benign keratosis. Total 10,015 images were collected from Human Against Machine dataset (HAM10000) available Kaggle, Even though shown best results many hyper-parameters that are required build difficult choose. If chosen doesn’t show good results, should be trained again with other set hyper-parameter values. To avoid circumstance, it done optimization. A base initially created without any technique, so performance by method can compared analysed. The first provided accuracy 73.34%, whereas improvement 77.17%.