Towards quantitative methods to assess network generative models

TL;DR

This paper proposes a novel method for evaluating network generative models by using graph classifiers to measure how well synthetic graphs resemble real networks in topological features.

Contribution

It introduces a new assessment approach that leverages graph classifiers to quantify the similarity between real and generated graphs, providing a performance metric for generative models.

Findings

Graph classifiers can effectively distinguish real from synthetic graphs.

The proposed method offers a quantitative measure of generative model quality.

This approach facilitates comparison of different network generative models.

Abstract

Assessing generative models is not an easy task. Generative models should synthesize graphs which are not replicates of real networks but show topological features similar to real graphs. We introduce an approach for assessing graph generative models using graph classifiers. The inability of an established graph classifier for distinguishing real and synthesized graphs could be considered as a performance measurement for graph generators.