Lets dig a little deep into how we convert the output of our CNN into probability — Softmax; and the loss measure to guide our optimization — Cross Entropy