Sometimes in deep learning, architecture design and hyperparameter tuning pose substantial challenges. Using Auto-Keras, none of these is needed: We start a search procedure and extract the best-performing model. This post presents Auto-Keras in action on the well-known MNIST dataset.
