Shapley-Value-Based Graph Sparsification for GNN Inference

TL;DR

This paper introduces a Shapley-value-based graph sparsification method that improves GNN inference efficiency by effectively pruning edges while maintaining prediction accuracy, leveraging the theoretical robustness of Shapley values.

Contribution

The paper presents a novel graph sparsification technique using Shapley values, which better captures edge importance by considering positive and negative contributions, outperforming existing explainability methods.

Findings

Maintains predictive performance after sparsification

Reduces graph complexity significantly

Enhances interpretability and efficiency in GNN inference

Abstract

Graph sparsification is a key technique for improving inference efficiency in Graph Neural Networks by removing edges with minimal impact on predictions. GNN explainability methods generate local importance scores, which can be aggregated into global scores for graph sparsification. However, many explainability methods produce only non-negative scores, limiting their applicability for sparsification. In contrast, Shapley value based methods assign both positive and negative contributions to node predictions, offering a theoretically robust and fair allocation of importance by evaluating many subsets of graphs. Unlike gradient-based or perturbation-based explainers, Shapley values enable better pruning strategies that preserve influential edges while removing misleading or adversarial connections. Our approach shows that Shapley value-based graph sparsification maintains predictive performance while significantly reducing graph complexity, enhancing both interpretability and efficiency in GNN inference.