A Convolutional neural network for recognizing hand written digits. The script uses Matplotlib for visualizing layers, output, loss and accuracy.
Input - 28 x 28
Convolutional layer - 28 x 28 filters: 24
Max pooling layer - 14 x 14
Convolutional layer - 14 x 14 filters: 200
Max pooling layer - 7 x 7
Convolutional layer - 7 x 7 filters: 20
Dropout layer - 0.8
Fully connected layer - 10 outputs (Softmax)
