Anomaly Detection In Cyber-Physical Systems Using Explainable Artificial Intelligence And Machine Learning
جاري التحميل...
التاريخ
عنوان الدورية
ردمد الدورية
عنوان المجلد
الناشر
Journal of Theoretical and Applied Information Technology
خلاصة
Cyber-Physical Systems (CPS) embrace integration between digital & physical components of production environments. Data analysis approaches operate on big data, which makes them somewhat limited in industrial applications. Not all of anomaly detection techniques are applicable in ensuring security of CPSs. These techniques face huge volumes of data and require domain-specific knowledge, which necessitates the invention of solutions that integrate advanced AI technologies and models. This paper utilizes Explainable Artificial Intelligence (XAI) & Machine Learning (ML) approaches for detecting the anomalies in CPS. The proposed model improves our understanding of the complex phenomena in CPSs by analyzing the extracted features using feature engineering selection and detecting the outliers of each class labels. Hence, the main motivation of this paper is to scrutinize challenges and emerging trends in Anomaly Detection for CPSs. Furthermore, studying the outlier detection algorithms such as Angle-based Outlier Detection (ABOD) and Clustering Based Local Outlier Factor (CBLOF) to be compared with the proposed approach. Neither of ABOD nor CBLOF succeeds in distinguishing the outlier class. Therefore, the proposed approach attempts to handle the outlier detection by using feature engineering and XAI approaches. Moreover, ML based Random Forest (RF) achieves better results than Support Vector Machine (SVM), Naïve Bayes (NB), and multi-layer perceptron (MLP).