Hidden Layer
A middle layer of neurons between the input and the output, where the learning happens. Each neuron acts as a feature detector and combines raw inputs into new, more abstract features.
When to use
Lecture 8's example: a hidden neuron can learn a "rainy evening" feature that has no column of its own. In PS 07, hidden_layer_sizes=(32,) means one hidden layer of 32 neurons.
See Lec 8.