Problem Set 13
This is to be completed by February 1st, 2018.
Exercises
* Complete the lesson:
a. Python Data Science Toolbox (Part II)

For a logistic regressor (multiclass ending in softmax) write down the update rules for gradient descent.

For a two layer perceptron ending in softmax with intermediate relu nonlinearity write down the update rules for gradient descent.
Python Lab
 Build a two layer perceptron (choose your nonlinearity) in numpy for a multiclass classification problem and test it on MNIST.
 Build a MLP in Keras and test it on MNIST.