Apache Spark Deep Learning Cookbook

How it works...

The neuron is the basic structural element of any neural network. A neuron can be thought of as a simple mathematical function or operator that operates on the input flowing through it to produce an output flowing out of it. The inputs to a neuron are multiplied by the node's weight matrix, summed over all the inputs, translated, and passed through an activation function. These are basically matrix operations in mathematics as described here:

1. The computational graph representation of a neuron is shown in the preceding diagram (b).
2. The transfer function for a single neuron or node is written as follows:

 

Here, x _i is the input ...