Create cat/dog classification tutorial 🐱 🐶

[ghost]

Summary

MNIST is one of the tutorials we have, but it's pretty simple. We shall add a new tutorial which has a bigger architecture; it'll help students to get more off of it, in terms of tuning and changing hyperparameters, as well as layers to visualize and feel the performance variations.

Intended Outcome

This tutorial will run a CNN network to classify photos into dogs or cats with an accuracy around 90%-95% or better.

How will it work?

It will run in the same way as MNIST tutorial is currently running. There shall be explanations alongside the code.