Explainable Anomaly Detection Framework for Cybersecurity in Smart Manufacturing  
Author Thi Hien Nguyen

 
Co-Author(s) Kim Duc Tran; Kim Phuc Tran

 
Abstract Anomaly detection is a major problem in smart manufacturing. Facing this problem, many solutions based on artificial intelligence have been researched and applied. In this work, we propose an unsupervised machine learning approach, Long short-term memory Autoencoders Elliptic Envelope (LSTM Autoencoders EE) methods, to handle anomaly detection. Based on specific data (the Gas Pipeline dataset), we present a table comparing the performance of the proposed method with other baseline algorithms. The experimental results on this data show that our proposed method provides the best performance in measure "Recall", this helps reduce risk of False Negatives during a prediction and therefore, reduce cost for manufacturer.

 
Keywords eXplainable Anomaly Detection (XAD), XAI, LSTM Autoencoder, Elliptic Envelope, Anomaly Detection
   
    Article #:  DSBFI25-1
 
Proceedings of 3rd ISSAT International Conference on Data Science in Business, Finance and Industry
January 6-8, 2025 - Da Nang, Vietnam