Beyond Fidelity: Explaining Vulnerability Localization of Learning-based Detectors

Abstract

Vulnerability detectors based on deep learning (DL) models have proven their effectiveness in recent years. However, the shroud of opacity surrounding the decision-making process of these detectors makes it difficult for security analysts to comprehend. To address this, various explanation approaches have been proposed to explain the predictions by highlighting important features, which have been demonstrated effective in other domains such as computer vision and natural language processing. Unfortunately, an in-depth evaluation of vulnerability-critical features, such as fine-grained vulnerability-related code lines, learned and understood by these explanation approaches remains lacking. In this study, we first evaluate the performance of ten explanation approaches for vulnerability detectors based on graph and sequence representations, measured by two quantitative metrics including fidelity and vulnerability line coverage rate. Our results show that fidelity alone is not sufficient for evaluating these approaches, as fidelity incurs significant fluctuations across different datasets and detectors. We subsequently check the precision of the vulnerability-related code lines reported by the explanation approaches, and find poor accuracy in this task among all of them. This can be attributed to the inefficiency of explainers in selecting important features and the presence of irrelevant artifacts learned by DL-based detectors.

arXiv commentAccepted by Tosem

References in corpus (11)

• Grad-CAM++: Improved Visual Explanations for Deep Convolutional Networks
• Interpretable machine learning: definitions, methods, and applications
• SySeVR: A Framework for Using Deep Learning to Detect Software Vulnerabilities
• Vulnerability Detection with Fine-grained Interpretations
• XGNN: Towards Model-Level Explanations of Graph Neural Networks
• Higher-Order Explanations of Graph Neural Networks via Relevant Walks
• VulDeeLocator: A Deep Learning-based Fine-grained Vulnerability Detector
• Post-hoc Interpretability for Neural NLP: A Survey
• Predicting Defective Lines Using a Model-Agnostic Technique
• Understanding Neural Code Intelligence Through Program Simplification
• Syntax-Guided Program Reduction for Understanding Neural Code Intelligence Models