As discussed in Chapter 6, Unsupervised Machine Learning Algorithms, a hyperparameter is a parameter whose value is chosen before the learning process starts. We start with common-sense values and then try to optimize them later. For neural networks, the important hyperparameters are these:
- The activation function
- The learning rate
- The number of hidden layers
- The number of neurons in each hidden layer
Let's look into how we can define a model using Keras.