Visiting Distant Neighbors in Graph Convolutional Networks

TL;DR

This paper enhances graph convolutional networks by incorporating higher-order neighbors, improving performance especially with limited labeled data, demonstrated through experiments on citation datasets.

Contribution

Introduces a higher-order neighbor visiting approach in GCNs, extending the model's receptive field for better node representations.

Findings

Higher-order neighbor visiting improves accuracy

Performance gains are significant with limited labeled data

Outperforms original GCN on citation datasets

Abstract

We extend the graph convolutional network method for deep learning on graph data to higher order in terms of neighboring nodes. In order to construct representations for a node in a graph, in addition to the features of the node and its immediate neighboring nodes, we also include more distant nodes in the calculations. In experimenting with a number of publicly available citation graph datasets, we show that this higher order neighbor visiting pays off by outperforming the original model especially when we have a limited number of available labeled data points for the training of the model.