Programming Machine Learning

Making It Deep

Let’s add one more layer to our three-layered neural network:

Each of the four layers in this deeper network comes with its own set of weights and its own activation function. The number of nodes in the new layer is yet one more hyperparameter, and we should be ready to tune it if we want the best results. To begin with, I set this value at 30.

Let’s turn this plan into code. That’s where using Keras really pays off, as adding this layer is a matter of adding one line of code to the model:

16_deeper/network_deep.py

model = Sequential()
model.add(Dense(100, activation='sigmoid'))
model.add(Dense(30, activation='sigmoid'))