Illuminati: Towards Explaining Graph Neural Networks for Cybersecurity Analysis

TL;DR

This paper introduces Illuminati, a comprehensive explanation framework for GNNs in cybersecurity, capable of identifying key graph components contributing to predictions without prior GNN knowledge, improving interpretability and accuracy.

Contribution

Illuminati is a novel, domain-agnostic explanation framework that enhances interpretability of GNNs in cybersecurity applications, outperforming existing methods in accuracy and clarity.

Findings

Achieves 87.6% subgraph retention of original prediction

Outperforms state-of-the-art explanation methods by 10.3%

Provides explanations easily understood by domain experts

Abstract

Graph neural networks (GNNs) have been utilized to create multi-layer graph models for a number of cybersecurity applications from fraud detection to software vulnerability analysis. Unfortunately, like traditional neural networks, GNNs also suffer from a lack of transparency, that is, it is challenging to interpret the model predictions. Prior works focused on specific factor explanations for a GNN model. In this work, we have designed and implemented Illuminati, a comprehensive and accurate explanation framework for cybersecurity applications using GNN models. Given a graph and a pre-trained GNN model, Illuminati is able to identify the important nodes, edges, and attributes that are contributing to the prediction while requiring no prior knowledge of GNN models. We evaluate Illuminati in two cybersecurity applications, i.e., code vulnerability detection and smart contract vulnerability detection. The experiments show that Illuminati achieves more accurate explanation results than state-of-the-art methods, specifically, 87.6% of subgraphs identified by Illuminati are able to retain their original prediction, an improvement of 10.3% over others at 77.3%. Furthermore, the explanation of Illuminati can be easily understood by the domain experts, suggesting the significant usefulness for the development of cybersecurity applications.