Neural Networks and Graphs
The machine learning landscape of the last decade has seen the rise and explosion of a particular type of model that is extremely popular nowadays, and whose name is becoming very familiar even to non-technical people and practitioners: artificial neural networks (ANNs). Their versatility and potency have resulted in widespread adoption globally, including in the graph domain. Several frameworks have been developed to support their study, use, and development.
Although the first attempts to train ANNs date back to the early 1980s (with the seminal work of Paul Werbos and Geoffrey Hinton), their rise and success has come around only recently, thanks to the advances in computing power (via CPUs but mostly thanks to the highly efficient parallelization of computation enabled by GPUs) as well as the availability of large datasets. ANNs are in fact very general models, able to virtually learn any function, but as such, they need to be trained on vast amounts...
