CAE_AD：基于卷积自编码器的无监督时间序列异常检测方法
CAE_AD: Unsupervised Time Series Anomaly Detection Method Based on Convolutional Autoencoder
DOI: 10.12677/JISP.2024.131003, PDF,   
作者: 韩广阳：北京邮电大学计算机学院(国家示范性软件学院)，北京；牛少彰：北京邮电大学计算机学院(国家示范性软件学院)，北京；东南数字经济发展研究院，浙江 衢州；王茂森, 安洪旭：东南数字经济发展研究院，浙江 衢州；史成洁：中国科学院信息工程研究所，北京
关键词: 时间序列；异常检测；对抗训练；卷积自编码器；Time Series； Abnormal Detection； Adversarial Training； Convolutional Autoencoder
摘要: 时间序列数据的有效异常检测对现代工业应用非常重要。然而，由于缺乏异常标签、数据高波动性、训练不稳定，导致建立一个能够准确地进行异常检测的系统是一个具有挑战性的问题。尽管异常检测的深度学习方法最近有所发展，但其中只有少数能够应对所有这些挑战。本文提出了CAE_AD，这是一种基于卷积自编码器(CAE)的无监督异常检测模型。为了尽量地放大异常，避免错过异常，笔者引入了两阶段的对抗训练。同时，为了提高训练稳定性，笔者引入了第一阶段的重建误差以作为第二阶段卷积自编码器的输入。笔者将CAE_AD与先进的时间序列异常检测方法在多个数据集上进行了比较。实验结果表明，本文提出的模型性能优于这些对比方法。在SMAP数据集上，相比于其他模型，CAE_AD模型的f1领先了4%，Precision领先了8%。
Abstract: Effective anomaly detection of time series data is crucial for modern industrial applications. How-ever, due to the lack of anomaly labels, high data volatility, and unstable training, establishing a system that can accurately detect anomalies is a challenging problem. Although deep learning methods for anomaly detection have recently developed, only a few of them can address all of these challenges. This article proposes CAE_AD, which is an unsupervised anomaly detection model based on convolutional autoencoder (CAE). In order to maximize the amplification of anomalies and avoid missing them, the author introduced two-stage adversarial training. Meanwhile, in order to improve training stability, the author introduced the reconstruction error from the first stage as the input for the convolutional autoencoder in the second stage. The author compared CAE_AD with advanced time series anomaly detection methods on multiple data sets. The experimental results show that the model proposed in this article performs better than these comparison methods. On the SMAP dataset, compared to other models, the CAE_AD model has a 4% lead in f1 and an 8% lead in Precision.
文章引用：韩广阳, 牛少彰, 王茂森, 史成洁, 安洪旭. CAE_AD：基于卷积自编码器的无监督时间序列异常检测方法[J]. 图像与信号处理, 2024, 13(1): 21-32. https://doi.org/10.12677/JISP.2024.131003

参考文献

[1] Huang, T., Zhu, Y., Zhang, Q., et al. (2013) Anlof-Based Adaptive Anomaly Detection Scheme for Cloud Computing. 2013 IEEE 37th Annual Computer Software and Applications Conference Workshops, Japan, 22-26 July 2013, 206-211. [Google Scholar] [CrossRef] 
[2] Zhang, L., Chen, Y. and Liao, S. (2018) Algorithm Optimiza-tion of Anomaly Detection Based on Data Mining. 2018 10th International Conference on Measuring Technology and Mechatronics Automation (ICMTMA), Changsha, 10-11 February 2018, 402-404. [Google Scholar] [CrossRef] 
[3] Çelik, M., Dadaşer-Çelik, F. and Dokuz, A.Ş. (2011) Anomaly Detection in Temperature Data Using DBSCAN Algorithm. 2011 International Symposium on Innovations in Intelligent Systems and Applications, Istanbul, 15-18 June 2011, 91-95. [Google Scholar] [CrossRef] 
[4] He, Z., Xu, X. and Deng, S. (2003) Discovering Clus-ter-Based Local Outliers. Pattern Recognition Letters, 24, 1641-1650. [Google Scholar] [CrossRef] 
[5] Su, M.Y. (2011) Real-Time Anomaly Detection Systems for Denial-of-Service Attacks by Weighted K-Nearest-Neighbor Classifiers. Expert Systems with Applications, 38, 3492-3498. [Google Scholar] [CrossRef] 
[6] Münz, G., Li, S. and Carle, G. (2007) Traffic Anom-aly Detection Using K-Means Clustering. Giitg Workshop Mmbnet, 7.
[7] Liu, F.T., Ting, K.M. and Zhou, Z.H. (2012) Isolation-Based Anomaly Detection. ACM Transactions on Knowledge Discovery from Data (TKDD), 6, 1-39. [Google Scholar] [CrossRef] 
[8] Muniyandi, A.P., Rajeswari, R. and Rajaram, R. (2012) Network Anomaly Detection by Cascading K-Means Clustering and C4.5 Decision Tree Algorithm. Procedia Engineering, 30, 174-182. [Google Scholar] [CrossRef] 
[9] Nanduri, A. and Sherry, L. (2016) Anomaly Detection in Air-craft Data Using Recurrent Neural Networks (RNN). 2016 IEEE Integrated Communications Navigation and Surveil-lance (ICNS), Herndon, VA, 19-21 April 2016, 5C2-1-5C2-8. [Google Scholar] [CrossRef] 
[10] Li, D., Chen, D., Jin, B., et al. (2019) MAD-GAN: Multi-variate Anomaly Detection for Time Series Data with Generative Adversarial Networks. International Conference on Ar-tificial Neural Networks, Springer International Publishing, Cham, 703-716. [Google Scholar] [CrossRef] 
[11] Hundman, K., Constantinou, V., Laporte, C., et al. (2018) De-tecting Spacecraft Anomalies Using LSTMS and Nonparametric Dynamic Thresholding. Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, London, 19-23 August 2018, 387-395. [Google Scholar] [CrossRef] 
[12] Su, Y., Zhao, Y., Niu, C., et al. (2019) Robust Anomaly Detection for Multivariate Time Series through Stochastic Recurrent Neural Network. Proceedings of the 25th ACM SIGKDD In-ternational Conference on Knowledge Discovery & Data Mining, Anchorage, AK, 4-8 August 2019, 2828-2837. [Google Scholar] [CrossRef] 
[13] Qi, S. (2018) Deep Autoencoding Gaussian Mixture Model for Unsupervised Anomaly Detection. International Conference on Learning Representations.
[14] Audibert, J., Michiardi, P., Guyard, F., et al. (2020) Usad: Unsupervised Anomaly Detection on Multivariate Time Series. Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, Virtual Event, CA, 6-10 July 2020, 3395-3404. [Google Scholar] [CrossRef] 
[15] Siffer, A., Fouque, P.A., Termier, A., et al. (2017) Anomaly Detection in Streams with Extreme Value Theory. Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Halifax, 13-17 August 2017, 1067-1075. [Google Scholar] [CrossRef]