Explainable android malware detection and malicious code localization using graph attention

doi:10.1016/j.jisa.2026.104385

Explainable android malware detection and malicious code localization using graph attention

2026-02-04

0

PRE

AI

DOI：10.1016/j.jisa.2026.104385

Original

Original request for help

Save

Abstract

En 中文

• Proposes a novel approach for automatic malicious code localization at both class and fine-grained method levels. • Integrates Graph Neural Networks and attention mechanisms to provide interpretable and explainable malware analysis. • Demonstrates robustness by evaluating on both synthetic and real-world malware datasets. • Significantly reduces manual analysis effort by automating the identification of malicious code segments. • Achieves a high recall rate of 97.27% and an F1-score of 95.30% at class-level and 96.02% recall with an F1-score of 92.68% at method-level localization.

Keywords:

malicious code localization

graph neural networks

attention mechanisms

android malware detection

explainable ai

Explainable android malware detection and malicious code localization using graph attention

Abstract

Journal

Organization