Improvements Chapter 01

1. diagram: samples vs. features generic matrix w/ terminology + add same matrix as example for MNIST and text classification
2. explain params (what learn) vs. hyper-params (how learn)