Machine Learning With Go

Nodes in the network

The nodes, or neurons, of our neural network have a relatively simple functionality in and of themselves. Each neuron will take in one or more values (x₁, x₂, and so on), combine these values according to an activation function, and produce a single output. The following is the output pictured:

How exactly should we combine the inputs to get the output? Well, we need a method to combine the inputs that is adjustable (such that we can train the model), and we have already seen that combining variables using coefficients and an intercept is one trainable way to combine inputs. Just think back to Chapter 4, Regression. In ...