Investigating Semi-supervised Time Series Anomaly Detection Using AW-SNBAE Model

Mingyuan Guo

doi:10.62051/ijcsit.v4n2.19

Authors

Mingyuan Guo

DOI:

https://doi.org/10.62051/ijcsit.v4n2.19

Keywords:

Anomaly detection, Time series analysis, Adaptive wavelet transform, N-Beats

Abstract

In time series anomaly detection, most of the existing algorithms have difficulty in balancing the precision and dependence on labeled data. To solve this problem, a semi-supervised model called Adaptive Wavelet Sparse N-Beats AutoEncoder (AW-SNBAE) is proposed. This model has two main innovations. Firstly, an adaptive wavelet transform module is designed, which adaptively learns the wavelet function based on the global information of the time series, removes noise, and captures features of short-term and long-term changes. Secondly, the proposed Sparse N-Beats Auto Encoder (SNBAE) uses N-Beats as a base, and then, combines residual enhancement and sparse regularization strategies to construct the encoder and the decoder, respectively. Experimental results show that the AW-SNBAE model improves F1 scores by 8% in univariate time series anomaly detection and by 2% on average in multivariate detection. These results demonstrate that the model is able to exhibit excellent detection capability despite limited labeled data.

Downloads

Download data is not yet available.

References

[1] Kozitsin V, Katser I, Lakontsev D. Online forecasting and anomaly detection based on the ARIMA model [J]. Applied Sciences, 2021, 11(7): 3194.

[2] Hosseinzadeh M, Rahmani A M, Vo B, et al. Improving security using SVM-based anomaly detection: issues and challenges [J]. Soft Computing, 2021, 25(4): 3195-3223.

[3] Jiang J, Chen J, Gu T, et al. Anomaly detection with graph convolutional networks for insider threat and fraud detection[C]//MILCOM 2019-2019 IEEE Military Communications Conference (MILCOM). IEEE, 2019: 109-114.

[4] Zong B, Song Q, Min M R, et al. Deep autoencoding gaussian mixture model for unsupervised anomaly detection[C]//International conference on learning representations. 2018.

[5] Mao J, Hu Y, Jiang D, et al. CBFS: a clustering-based feature selection mechanism for network anomaly detection [J]. IEEE Access, 2020, 8: 116216-116225.

[6] Deng A, Hooi B. Graph neural network-based anomaly detection in multivariate time series[C]//Proceedings of the AAAI conference on artificial intelligence. 2021, 35(5): 4027-4035.

[7] Pelati A, Meo M, Dini P. Traffic anomaly detection using deep semi-supervised learning at the mobile edge [J]. IEEE Transactions on Vehicular Technology, 2022, 71(8): 8919-8932.

[8] Yu Q, Kavitha M S, Kurita T. Mixture of experts with convolutional and variational autoencoders for anomaly detection [J]. Applied Intelligence, 2021, 51(6): 3241-3254.

[9] Kumagai A, Iwata T, Fujiwara Y. Semi-supervised anomaly detection on attributed graphs[C]//2021 International Joint Conference on Neural Networks (IJCNN). IEEE, 2021: 1-8.

[10] Li P, Pei Y, Li J. A comprehensive survey on design and application of autoencoder in deep learning [J]. Applied Soft Computing, 2023, 138: 110176.

[11] Oreshkin B N, Carpov D, Chapados N, et al. N-BEATS: Neural basis expansion analysis for interpreTable time series forecasting [J]. arxiv preprint arxiv:1905.10437, 2019.

[12] Zhang D, Zhang D. Wavelet transform [J]. Fundamentals of image data mining: Analysis, Features, Classification and Retrieval, 2019: 35-44.

[13] Liu X, Liu H, Guo Q, et al. Adaptive wavelet transform model for time series data prediction [J]. Soft Computing, 2020, 24(8): 5877-5884.

[14] Kim J, Kang H, Kang P. Time-series anomaly detection with stacked Transformer representations and 1D convolutional network [J]. Engineering Applications of Artificial Intelligence, 2023, 120: 105964.

[15] Belghazi M I, Baratin A, Rajeshwar S, et al. Mutual information neural estimation[C]//International conference on machine learning. PMLR, 2018: 531-540.

[16] Greenacre M, Groenen P J F, Hastie T, et al. Principal component analysis [J]. Nature Reviews Methods Primers, 2022, 2(1): 100.

[17] Sauvanaud C, Kaâniche M, Kanoun K, et al. Anomaly detection and diagnosis for cloud services: Practical experiments and lessons learned [J]. Journal of Systems and Software, 2018, 139: 84-106.

[18] Li W, Feng C, Chen T, et al. Robust learning of deep time series anomaly detection models with contaminated training data [J]. arxiv preprint arxiv:2208.01841, 2022.

[19] Skolkovo Institute of Science and Technology. (n.d.). SKAB: Skoltech anomaly benchmark. GitHub. https://github.com/waico/SKAB

[20] **a X, Pan X, Li N, et al. GAN-based anomaly detection: A review [J]. Neurocomputing, 2022, 493: 497-535.

[21] Gouda W, Tahir S, Alanazi S, et al. Unsupervised outlier detection in IOT using deep VAE [J]. Sensors, 2022, 22(17): 6617.