What is a neural network?
Notice: This is strictly the text from the attachment. There are pictures that will help to illustrate the ideas in this document.
At its core, a neural network is designed to be a computer simulation of the human brain. It is represented by a labeled, directed graph structure, where nodes perform some simple computations. Each node represents a neuron, and each directed edge represents a weighted connection between two nodes.
A very simple example of a neural network is the following AND-gate.
In this example, nodes x and y are the inputs to node z. w_1 and w_2 are the weights on the connection. In order for this to operate as an AND-gate, we need to set w_1 and w_2 to 1. This means that the input from x will be multiplied by the weight w_1, and the input from y will be multiplied by the weight w_2. We also need to define the operation in node z. If we say that z calculates the product of its inputs, that is, z calculates (x*w_1)(x*w_2), we will have an AND-gate. Why is this true? We can examine the truth ...
This job explores neural networks.