Mole Scan A deep neural network training on the HAM10000 Skin cancer MNIST dataset, split 7:3 for training:testing. Achieved 75% accuracy without a convolutional layer using grayscale images only. Created for McGill Physics Hackathon 2019.