Programming Machine Learning

Chapter 19Beyond Vanilla Networks

At this point in the book, you’re probably fairly comfortable with the structure of a neural network. Hold on tight, because we’re about to shake things up a little and peek at neural networks that stray beyond that familiar structure.

All the networks that we’ve seen so far share the same architecture: a sequence of dense layers, where by “dense” I mean that each node in a layer is connected to all the nodes in the neighboring layers. That blueprint is also called a fully connected neural network—and as it turns out, it comes with a drastic limitation. Whether it’s classifying images, parsing a product review, or predicting traffic congestion, a fully connected network treats all data the same, as an indistinct ...