In this issue:

• we discuss building blocks and types of GNN architectures; 

• we explain GraphWorld, which provides insights about how to test GNNs; 

• we explore Spektral, a library for building GNNs in Keras and TensorFlow. 

Enjoy the learning!  

💡 ML Concept of the Day: Building Blocks and Types of GNN Architectures  

In this installment of our series about graph neural networks (GNNs), we would like to explore the key components of any GNN architecture. As explored in this series, the key idea of a GNN is to create graph structures that represent knowledge of a specific task and environment. The architectures of GNNs can be very dissimilar but, in general, there are three key components that are present in most of these neural networks:  ...