Learnings on machine learning.

Deriving canonical correlation analysis

In which we explore CCA in detail

The Gumbel-max trick

A funky way of sampling from categorical distributions