Graph Pattern Mining and Learning through User-defined Relations (Extended Version)

TL;DR

This paper introduces R-GPM, a scalable parallel framework for graph pattern mining using user-defined relations, which employs MCMC sampling and optimizations to efficiently estimate pattern statistics with theoretical guarantees.

Contribution

The work presents a novel parallel framework for graph pattern mining that generalizes traditional methods through user-defined relations and offers scalable, accurate estimators with theoretical support.

Findings

R-GPM achieves up to 3-orders-of-magnitude reduction in computational costs.

Provides near-linear speedups on 44-core systems.

Demonstrates effectiveness in deep graph neural network optimization and motif counting.

Abstract

In this work we propose R-GPM, a parallel computing framework for graph pattern mining (GPM) through a user-defined subgraph relation. More specifically, we enable the computation of statistics of patterns through their subgraph classes, generalizing traditional GPM methods. R-GPM provides efficient estimators for these statistics by employing a MCMC sampling algorithm combined with several optimizations. We provide both theoretical guarantees and empirical evaluations of our estimators in application scenarios such as stochastic optimization of deep high-order graph neural network models and pattern (motif) counting. We also propose and evaluate optimizations that enable improvements of our estimators accuracy, while reducing their computational costs in up to 3-orders-of-magnitude. Finally,we show that R-GPM is scalable, providing near-linear speedups on 44 cores in all of our tests.