The application of deep learning on fast skin cancer diagnosis

Nowadays, health problems have been more and more important. In order to solve the difficulties of diagnosing skin cancer quickly and precisely, we applied a state-of-the-art deep convolutional neural network (CNN) architecture with transfer learning, to help diagnose various types of skin cancers. Combined with proper data augmentation and normalization, the designed model achieves state-of-the-art result, including f1-score to be as high as 0.83, which demonstrates a promising application of the deep neural network on visual diagnosis of skin illness.