A Generic Graph Sparsification Framework using Deep Reinforcement Learning

TL;DR

This paper introduces SparRL, a deep reinforcement learning framework for graph sparsification that adapts to various objectives and outperforms existing methods in preserving graph properties.

Contribution

The paper presents the first generic, flexible, and scalable deep RL-based framework for graph sparsification, addressing limitations of sampling-based methods.

Findings

SparRL outperforms existing sparsification methods across multiple metrics.

The framework adapts effectively to different reduction objectives.

Graph size-independent complexity is achieved with SparRL.

Abstract

The interconnectedness and interdependence of modern graphs are growing ever more complex, causing enormous resources for processing, storage, communication, and decision-making of these graphs. In this work, we focus on the task graph sparsification: an edge-reduced graph of a similar structure to the original graph is produced while various user-defined graph metrics are largely preserved. Existing graph sparsification methods are mostly sampling-based, which introduce high computation complexity in general and lack of flexibility for a different reduction objective. We present SparRL, the first generic and effective graph sparsification framework enabled by deep reinforcement learning. SparRL can easily adapt to different reduction goals and promise graph-size-independent complexity. Extensive experiments show that SparRL outperforms all prevailing sparsification methods in producing high-quality sparsified graphs concerning a variety of objectives.