Learning Graph Embedding With Adversarial Training Methods

TLDR

Graph embedding transforms graph data into vector representations to support tasks such as link prediction and clustering, yet most methods focus on structural preservation or reconstruction errors while neglecting the distribution of latent codes, which can degrade representation quality. This work introduces an adversarially regularized framework to learn graph embeddings. The framework employs a graph convolutional network encoder to embed topology and node attributes, a decoder to reconstruct the graph, and adversarial training to align latent codes with a Gaussian or uniform prior, yielding two variants—ARGA and ARVGA—along with additional design variations. Experiments comparing 12 link‑prediction and 20 clustering algorithms demonstrate that the proposed methods outperform existing approaches.

Abstract

Graph embedding aims to transfer a graph into vectors to facilitate subsequent graph-analytics tasks like link prediction and graph clustering. Most approaches on graph embedding focus on preserving the graph structure or minimizing the reconstruction errors for graph data. They have mostly overlooked the embedding distribution of the latent codes, which unfortunately may lead to inferior representation in many cases. In this article, we present a novel adversarially regularized framework for graph embedding. By employing the graph convolutional network as an encoder, our framework embeds the topological information and node content into a vector representation, from which a graph decoder is further built to reconstruct the input graph. The adversarial training principle is applied to enforce our latent codes to match a prior Gaussian or uniform distribution. Based on this framework, we derive two variants of the adversarial models, the adversarially regularized graph autoencoder (ARGA) and its variational version, and adversarially regularized variational graph autoencoder (ARVGA), to learn the graph embedding effectively. We also exploit other potential variations of ARGA and ARVGA to get a deeper understanding of our designs. Experimental results that compared 12 algorithms for link prediction and 20 algorithms for graph clustering validate our solutions.