communities, modules, community detection, graph neural networks

Detecting communities or classifying nodes is an important task which may be accomplished via different (well explored) approaches. The most popular are spectral and probabilistic methods. With the recent expansion of data-driven methods which often replace hand-designed heuristics, new possibilities for community detection have appeared. One of them is a new branch of neural networks. Namely Graph Neural Networks (GNNs) which are an extension of Convolutional Neural Networks to graph-structured data. This contribution aims to present GNNs, compare them to traditional methods and show an example of implementation.