Implementing a Three-Layer Neural Network

Now, let's implement a "practical" neural network. Here, we will implement the process from its input to its output (a process in the forward direction) in the three-layer neural network shown in Figure 3.15. We will use NumPy's multidimensional arrays (as described in the previous section) for implementation. By making good use of NumPy arrays, you can write some short code for a forward process in the neural network.

Examining the Symbols

Here, we will use symbols such as and to explain the processes performed in the neural network. They may seem a little complicated. You can skim through this section because the symbols are only used here:

Figure 3.15: A three-layer neural network consisting of two neurons in the input layer (layer 0), three neurons in the first hidden layer (layer 1), two neurons in the second hidden layer (layer 2), and two neurons in the output layer (layer 3)

Note

...