Detecting Anomalies in Time-Series Data using Unsupervised Learning and Analysis on Infrequent
1
조회
0
다운로드
2
인용
0
참고문헌
키워드
초록
We propose a framework called Stacked Gated Recurrent Unit - Infrequent Residual Analysis (SG-IRA) thatdetects anomalies in time-series data that can be trained on streams of raw sensor data without any pre-labeleddataset. To enable such unsupervised learning, SG-IRA includes an estimation model that uses a stacked GatedRecurrent Unit (GRU) structure and an analysis method that detects anomalies based on the difference betweenthe estimated value and the actual measurement (residual). SG-IRA’s residual analysis method dynamically adaptsthe detection threshold from the population using frequency analysis, unlike the baseline model that relies on aconstant threshold. In this paper, SG-IRA is evaluated using the industrial control systems (ICS) datasets. SG-IRAimproves the detection performance (F1 score) by 5.9% compared to the baseline model.
Abstract
We propose a framework called Stacked Gated Recurrent Unit - Infrequent Residual Analysis (SG-IRA) thatdetects anomalies in time-series data that can be trained on streams of raw sensor data without any pre-labeleddataset. To enable such unsupervised learning, SG-IRA includes an estimation model that uses a stacked GatedRecurrent Unit (GRU) structure and an analysis method that detects anomalies based on the difference betweenthe estimated value and the actual measurement (residual). SG-IRA’s residual analysis method dynamically adaptsthe detection threshold from the population using frequency analysis, unlike the baseline model that relies on aconstant threshold. In this paper, SG-IRA is evaluated using the industrial control systems (ICS) datasets. SG-IRAimproves the detection performance (F1 score) by 5.9% compared to the baseline model.
참고문헌 (6)
- J. Chung, "Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling", NIPS 2014Workshop on Deep Learning (2014)
- "Industrial Control Systems Security Threat Detection AI Competition"[URL]
- H. Shin, "HAI 1.0:HIL-based Augmented ICS Security Dataset", 13th USENIX Workshop on Cyber Security Experimentation and Test (CSET) (2020)
- W.-Hwang, "Time-Series Aware Precision and Recall for Anomaly Detection:Considering Variety of Detection Result and Addressing Ambiguous Labeling", Proc. of CIKM, pp.2241~2244 (2019)
- "eTaPR"[URL]
- X. Chen, "Multi-scale anomaly detection algorithm based on infrequent pattern of time series", Computational and Applied Mathematics, 214, pp.227~237 (2008)[DOI]