Implement a Handwritten Numbers classification.

Dataset: THE MNIST DATABASE of handwritten digits (Full Dataset) Use 60,000 train images of MNIST for Training and 10,000 test images for Testing (MNIST Dataset already separated to Train and Test sets).

Apply Convolutional Neural Networks (CNNs) for architecture.
CNN > ReLU > MaxPooling > CNN > ReLU > MaxPooling > CNN > ReLU > MaxPooling > fully connected layer > output