Global-Local Graph Neural Networks for Node-Classification

Moshe Eliasof, Eran Treister

Keywords: Graph Neural Networks, Deep Learning, Node classification

TL;DR: We propose a method to improve graph node-classification using global (label) information.

Abstract:

The task of graph node-classification is often approached using a \emph{local} Graph Neural Network (GNN), that learns only local information from the node input features and their adjacency. In this paper we propose to benefit from global and local information through the form of learning \emph{label}- and \emph{node}- features to improve node-classification accuracy. We therefore call our method Global-Local-GNN (GLGNN).
To learn proper label features, for each label, we maximize the similarity between its features and nodes features that belong to the label, while maximizing the distance between nodes that do not belong to the considered label. We then use the learnt label features to predict the node-classification map. We demonstrate our GLGNN using GCN and GAT as GNN backbones, and show that our GLGNN approach improves baseline performance on the node-classification task.

Type Of Submission: Extended abstract (max 4 main pages).

Agreement: Check this if you are okay with being contacted to participate in an anonymous survey.

Type Of Submission: Extended abstract.