Machine Learning With Go

Why do we expect this architecture to work?

Let's pause for a moment and think about why setting up a series of nodes in the preceding arrangement would allow us to predict something. As you can see, all we are doing is combining inputs over and over to produce some result. How can we expect this result to be useful in making binary classifications?

Well, what exactly do we mean when we define a binary classification problem, or any classification or regression problem, for that matter? We are basically saying that we expect there to be some rules or relationships that exist between a series of inputs (our attributes) and an output (our labels or response). In essence, we are saying that there exists some simple or complicated function that ...