Self-supervised Consensus Representation Learning for Attributed Graph

TL;DR

This paper introduces SCRL, a self-supervised learning framework that leverages both topology and feature graphs for attributed graph representation, improving semi-supervised node classification performance.

Contribution

The novel SCRL framework jointly learns from topology and feature graphs using self-supervision, which is a new approach in attributed graph representation learning.

Findings

SCRL outperforms state-of-the-art methods on citation and social network datasets.

SCRL is more efficient than comparable methods.

Self-supervised agreement maximization improves node embeddings.

Abstract

Attempting to fully exploit the rich information of topological structure and node features for attributed graph, we introduce self-supervised learning mechanism to graph representation learning and propose a novel Self-supervised Consensus Representation Learning (SCRL) framework. In contrast to most existing works that only explore one graph, our proposed SCRL method treats graph from two perspectives: topology graph and feature graph. We argue that their embeddings should share some common information, which could serve as a supervisory signal. Specifically, we construct the feature graph of node features via k-nearest neighbor algorithm. Then graph convolutional network (GCN) encoders extract features from two graphs respectively. Self-supervised loss is designed to maximize the agreement of the embeddings of the same node in the topology graph and the feature graph. Extensive experiments on real citation networks and social networks demonstrate the superiority of our proposed SCRL over the state-of-the-art methods on semi-supervised node classification task. Meanwhile, compared with its main competitors, SCRL is rather efficient.