In this section, we're going to create and train a support vector machine that will actually perform our digit classification.
In the very first example, we're going to use scikit-learn, and we're going to use what's called a support vector machine, which is a very powerful, very versatile classic machine learning technique that can learn all kinds of functions and all kinds of mappings from inputs to outputs. We're going to do classification, which is mapping inputs as an array of pixels, and in our case we're going to classify each input into one of ten classes, corresponding to ten digits. But we can classify different kinds of things as continuous ordered functions, which is called regression, and that can be useful, for example, if you want to extract position or an area of volume where it doesn...