New Frontiers in Graph Autoencoders: Joint Community Detection and Link Prediction

TL;DR

This paper introduces a novel community-preserving message passing scheme for graph autoencoders, enabling high-accuracy joint community detection and link prediction, outperforming existing methods on real-world graphs.

Contribution

It proposes Modularity-Aware GAE and VGAE with new training strategies for simultaneous community detection and link prediction, addressing a key gap in current graph autoencoder performance.

Findings

Effective joint community detection and link prediction demonstrated

Outperforms simpler methods like Louvain in experiments

Applicable to various real-world graph datasets

Abstract

Graph autoencoders (GAE) and variational graph autoencoders (VGAE) emerged as powerful methods for link prediction (LP). Their performances are less impressive on community detection (CD), where they are often outperformed by simpler alternatives such as the Louvain method. It is still unclear to what extent one can improve CD with GAE and VGAE, especially in the absence of node features. It is moreover uncertain whether one could do so while simultaneously preserving good performances on LP in a multi-task setting. In this workshop paper, summarizing results from our journal publication (Salha-Galvan et al. 2022), we show that jointly addressing these two tasks with high accuracy is possible. For this purpose, we introduce a community-preserving message passing scheme, doping our GAE and VGAE encoders by considering both the initial graph and Louvain-based prior communities when computing embedding spaces. Inspired by modularity-based clustering, we further propose novel training and optimization strategies specifically designed for joint LP and CD. We demonstrate the empirical effectiveness of our approach, referred to as Modularity-Aware GAE and VGAE, on various real-world graphs.