4. Problem 4: In this problem, we utilize Tensorflow and keras to train a convolutional NN for clas-

sification using the Fashion MNIST (M = 10) dataset ³. In this dataset, we have N = 60000

training data points and 10000 examples for testing where each one consists of a gray scale image

with F = 28 * 28 = 784 features (or pixels). The dataset is loaded and pre-processed using the

following code.