Rules Refine the Riddle: Global Explanation for Deep Learning-Based Anomaly Detection in Security Applications

• DOI: 10.1145/3658644.3670375
• Publication Date: 2024-12-09T17:19:20Z
• Authors: Dongqi Han; Zhiliang Wang; Ruitao Feng; Minghui Jin; Wenqi Chen; Kai Wang; Su Wang; Jiahai Yang; Xingang Shi; Xia Yin; Yang Liu
• Keywords: Interpretability; Anomaly (physics)

Deep learning (DL) based anomaly detection has shown great promise in the field of security due to its remarkable performance various tasks. However, issue poor interpretability DL models significantly impeded their deployment practical applications. Despite progress made existing studies on explanations, majority them focus providing local explanations for individual samples, neglecting global understanding model knowledge. Furthermore, most supervised fail apply different mechanisms.