Wip A Formal Model: The Statistical Learning Framework

15 Jul 2020 - pinaki

this is a work in progress…please view later.

Can we get some theoretical guarantees of how successful learning can be achieved in a relatively simplified setting?

In this blog, I shall answer the aforementioned question. Before starting, I must mention that I learnt about this concept from “Understanding Machine Learning: From Theory to Algorithms” by Shai Shalev-Shwartz and Shai Ben-David, Chapter 2 page 33.

So let’s get cracking…

Before diving deep into theory let us first define the learner’s input i.e. the basic statistical learning setting that the learner has access to:

Consider that the learner gets access to a finite set of papayas which it can taste and then decide whether it is 1 (tasty) or 0 (non-tasty). This finite set or the training data are often called training examples.

\[\nabla_\boldsymbol{x} J(\boldsymbol{x})\]