Size Matters: Large Graph Generation with HiGGs

Alex O. Davies,Nirav S. Ajmeri,Telmo M. Silva Filho

2023-11-07

Abstract:Large graphs are present in a variety of domains, including social networks, civil infrastructure, and the physical sciences to name a few. Graph generation is similarly widespread, with applications in drug discovery, network analysis and synthetic datasets among others. While GNN (Graph Neural Network) models have been applied in these domains their high in-memory costs restrict them to small graphs. Conversely less costly rule-based methods struggle to reproduce complex structures. We propose HIGGS (Hierarchical Generation of Graphs) as a model-agnostic framework of producing large graphs with realistic local structures. HIGGS uses GNN models with conditional generation capabilities to sample graphs in hierarchies of resolution. As a result HIGGS has the capacity to extend the scale of generated graphs from a given GNN model by quadratic order. As a demonstration we implement HIGGS using DiGress, a recent graph-diffusion model, including a novel edge-predictive-diffusion variant edge-DiGress. We use this implementation to generate categorically attributed graphs with tens of thousands of nodes. These HIGGS generated graphs are far larger than any previously produced using GNNs. Despite this jump in scale we demonstrate that the graphs produced by HIGGS are, on the local scale, more realistic than those from the rule-based model BTER.

Machine Learning,Artificial Intelligence,Social and Information Networks

What problem does this paper attempt to address?

The main objective of this paper is to address the problem of large-scale graph generation. Specifically, existing Graph Neural Network (GNN) models are limited by memory constraints when generating large-scale graphs and can only handle smaller graphs; rule-based methods, while cost-effective, struggle to reproduce complex structures. To tackle these issues, the authors propose a framework called HIGGS (Hierarchical Generation of Graphs). This framework leverages the conditional generation capabilities of GNN models to generate graphs progressively at different resolution levels and expands the scale of the generated graphs hierarchically, thereby enabling the generation of large-scale graphs with tens of thousands of nodes while maintaining the authenticity of local structures. Additionally, the HIGGS framework can generate large-scale graphs with node and edge attributes, which is not achievable with existing standalone GNN models. Compared to rule-based methods, HIGGS not only generates larger-scale graphs but also preserves the authenticity of local structures.

Size Matters: Large Graph Generation with HiGGs

HiGen: Hierarchical Graph Generative Networks

Hierarchical recurrent neural networks for graph generation

Scalable Deep Generative Modeling for Sparse Graphs

SaGess: Sampling Graph Denoising Diffusion Model for Scalable Graph Generation

Efficient and Scalable Graph Generation through Iterative Local Expansion

GraphMaker: Can Diffusion Models Generate Large Attributed Graphs?

Towards Efficient Large-Scale Graph Neural Network Computing.

Enhancing Size Generalization in Graph Neural Networks through Disentangled Representation Learning

On the Scalability of GNNs for Molecular Graphs

HYGENE: A Diffusion-based Hypergraph Generation Method

Scalable and Efficient Full-Graph GNN Training for Large Graphs

IGB: Addressing The Gaps In Labeling, Features, Heterogeneity, and Size of Public Graph Datasets for Deep Learning Research

Scalable Graph Neural Networks Via Bidirectional Propagation.

Large Generative Graph Models

Efficient Graph Neural Network Inference at Large Scale

DiGress: Discrete Denoising diffusion for graph generation

DistDGL: Distributed Graph Neural Network Training for Billion-Scale Graphs

Improving Equivariant Graph Neural Networks on Large Geometric Graphs Via Virtual Nodes Learning

InferTurbo: A Scalable System for Boosting Full-graph Inference of Graph Neural Network over Huge Graphs

Large-scale graph representation learning with very deep GNNs and self-supervision