In this section, we'll describe the Neural Graph Learning paradigm (NGL) (for more information, see Neural Graph Learning: Training Neural Networks Using Graphs at https://storage.googleapis.com/pub-tools-public-publication-data/pdf/bbd774a3c6f13f05bf754e09aa45e7aa6faa08a8.pdf), which makes it possible to augment training based on unstructured data with structured signals. More specifically, we'll discuss the neural structured learning framework (NSL) (for more information, go to https://www.tensorflow.org/neural_structured_learning/), which is based on TensorFlow 2.0 and implements these principles.
To understand how NGL works, we'll use the CORA dataset (https://relational.fit.cvut.cz/dataset/CORA), which consists ...
