Introduction to GNNs
As you will see shortly, the concepts introduced in the previous section can be quite naturally extended to deal with graphs.
GNNs are deep learning methods that work on graph-structured data. This family of methods is sometimes also referred to as geometric deep learning and is gaining increasing popularity in a variety of applications, including social network analysis and computer graphics. The underlying idea of GNNs is a natural extension of CNNs, which we just used in the previous example to process images and have achieved impressive results when dealing with regular Euclidean spaces, such as text (one-dimensional), images (two-dimensional), and videos (three-dimensional). As shown in the previous example, CNN layers combine inputs from their neighborhood, applying a static kernel that is swept throughout the entire space – that is, every single pixel making up the image. Moreover, a classic CNN consists of a sequence of layers and each layer...
