Figure 5: Comparison of GenStat architecture to selected graph generative models. This proof uses two properties of LDP: composability and immunity to post-processing [2]. Figure 6 illustrates the PGM of Randomized algorithms. The GGM parameters are a function of the perturbed graph statistics as learning input. The implementation can be easily extended to directed graphs. A statistics-based GGM that takes the degree sequence as sufficient statistics [5].

Neural Information Processing Systems http://nips.cc/

Why do many modern neural-network-based graph generative models fail to reproduce typical real-world network characteristics, such as high triangle density?

A wide range of models have been proposed for Graph Generative Models, necessitating effective methods to evaluate their quality. So far, most techniques use either traditional metrics based on subgraph counting, or the representations of randomly initialized Graph Neural Networks (GNNs). We propose using representations from constrastively trained GNNs, rather than random GNNs, and show this gives more reliable evaluation metrics. Neither traditional approaches nor GNN-based approaches dominate the other, however: we give examples of graphs that each approach is unable to distinguish. We demonstrate that Graph Substructure Networks (GSNs), which in a way combine both approaches, are better at distinguishing the distances between graph datasets.

Figure 5: Comparison of GenStat architecture to selected graph generative models. This proof uses two properties of LDP: composability and immunity to post-processing [2]. Figure 6 illustrates the PGM of Randomized algorithms. The GGM parameters are a function of the perturbed graph statistics as learning input. The implementation can be easily extended to directed graphs. A statistics-based GGM that takes the degree sequence as sufficient statistics [5].

We propose a new family of efficient and expressive deep generative models of graphs, called Graph Recurrent Attention Networks (GRANs).

Why do many modern neural-network-based graph generative models fail to reproduce typical real-world network characteristics, such as high triangle density?

Why do many modern neural-network-based graph generative models fail to reproduce typical real-world network characteristics, such as high triangle density?

A wide range of models have been proposed for Graph Generative Models, necessitating effective methods to evaluate their quality. So far, most techniques use either traditional metrics based on subgraph counting, or the representations of randomly initialized Graph Neural Networks (GNNs). We propose using representations from contrastively trained GNNs, rather than random GNNs, and show this gives more reliable evaluation metrics. Neither traditional approaches nor GNN-based approaches dominate the other, however: we give examples of graphs that each approach is unable to distinguish. We demonstrate that Graph Substructure Networks (GSNs), which in a way combine both approaches, are better at distinguishing the distances between graph datasets.