MNIST_digit_classification

In this jupyter notebook, I have used Tensorflow library to classify the digits in MNIST dataset. ---I have used ANN with two dense layer and one output layer, several activation functions and loss functions are used and comparison between them is drawn by calculating accuracy from each of them. All activation functions and loss function have received equally well accuracy. ---Lastly, the number of neurons are changed in the input layers while using the same combination of activation function and loss function. It is found that increasing the number of neurons in the hidden layers, improves the accuracy.