基于深度学习的异常检测模型综述

摘要/Abstract

摘要： 随着数据复杂度的激增，传统的异常检测方法不再适用，而深度学习可通过多层神经网络处理复杂数据并进行异常检测。通过对国内外研究成果的广泛调研，首先，介绍了异常检测及其分类，并将基于深度学习的异常检测模型分为基于生成模型、基于重构模型和基于单分类模型 3 类。其次，梳理了不同类别下深度异常检测模型的构造原理和发展脉络，并分析了不同模型的适用范围和优缺点；介绍了异常检测常用的数据集和评估指标，并对比分析了不同模型的性能。最后总结了深度异常检测面临的挑战，并对未来的研究方向进行了展望。

关键词:

深度学习, 异常检测, 机器学习, 自动编码器, 神经网络

Abstract:

With the proliferation of data complexity, traditional methods for anomaly detection are no longer applicable, while deep learning can process complex data and perform anomaly detection by multi-layer neural network. Through extensive research on domestic and international research results, firstly, anomaly detection and the types of detection methods are introduced. And deep learning based anomaly detection models are classified into three categories of generative-based model, reconstruction-based model and single classification-based model. Secondly,the construction principles and development context of deep anomaly detection models under different categories are sorted out, and the applicability, advantages and disadvantages of various models are analyzed. Then, the commonly used datasets and evaluation indicators for anomaly detection are introduced, and the performance of different models are compared and analyzed. Finally, the challenges faced by deep anomaly detection are summarized and the future research directions are prospected.

Key words:

deep learning, anomaly detection, machine learning, autoencoder, neural network

中图分类号:

TP301

周茂袁, 伍小双.

基于深度学习的异常检测模型综述

[J]. 中国民航大学学报, 2023, 41(4): 1-7.

ZHOU Maoyuan, WU Xiaoshuang.

Review of anomaly detection models based on deep learning

[J]. Journal of Civil Aviation University of China, 2023, 41(4): 1-7.

参考文献 45

[1]	王鑫, 张涛, 金映谷. 异常检测算法综述[J]. 现代计算机, 2020(30): 21-26.
[2]	卓琳, 赵厚宇, 詹思延. 异常检测方法及其应用综述[J]. 计算机应用研究, 2020, 37(S1): 9-15.
[3]	RADFORD A, METZ L, CHINTALA S. Unsupervised representation learning with deep convolutional generative adversarial networks[EB/OL].(2016-01-07) [2022-08-31]. https://doi.org/10.48550/arXiv.1511.06434.
[4]	ARJOVSKY M, CHINTALA S, BOTTOU L. Wasserstein generative adversarial networks[C]//The 34th International Conference on MachineLearning, August 6 -11, 2017, Sydney, NSW, Australia. ACM, 2017:214-223.
[5]	SCHLEGL T, SEEB魻CK P, WALDSTEIN S M, et al. Unsupervised anomaly detection with generative adversarial networks to guide marker discovery [C]//International Conference on Information Processing inMedical Imaging. Cham: Springer, 2017: 146-157.
[6]	DEECKE L, VANDERMEULEN R, RUFF L, et al. Image anomaly detection with generative adversarial networks[C]//Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Cham: Springer, 2019: 3-17.
[7]	SCHLEGL T, SEEB魻CK P, WALDSTEIN S M, et al. F-AnoGAN: fast unsupervised anomaly detection with generative adversarial networks[J]. Medical Image Analysis, 2019, 54: 30-44.
[8]	AKCAY S, ATAPOUR-ABARGHOUEI A, BRECKON T P. GANomaly: semi-supervised anomaly detection via adversarial training[C]//The 14th Asian Conference on Computer Vision, December 2-6, 2018, Perth, Australia. Cham: Springer, 2018: 622-637.
[9]	赵飏, 李晓, 马博, 等. 基于LSTM-GAN 的加油时序数据异常检测[J].计算机应用与软件, 2022, 39(7): 13-19.
[10]	AN J, CHO S. Variational autoencoder based anomaly detection using reconstruction probability[EB/OL]. (2015-11-27)[2022-08-31]. http://dm.snu.ac.kr/static/docs/TR/SNUDM-TR-2015-03.pdf.
[11]	SINHA S, GIFFARD-ROISIN S, KARBOU F, et al. Variational autoencoderanomaly-detection of avalanche deposits in satellite SAR imagery[C]//Proceedings of the 10th International Conference on Climate Informatics, September 22-25, 2020, Virtual, United Kingdom. ACM, 2020:113-119.
[12]	ZENO B, MATVEEV Y N, ALKHATIB B. Face validation based anomaly detection using variational autoencoder[J]. IOP Conference Series: Materials Science and Engineering, 2019, 618: 012011.
[13]	WASEEM F, MARTINEZ R P, WU C. Visual anomaly detection in video by variational autoencoder[EB/OL]. (2022-03-08) [2022-08-31]. https://arxiv.org/abs/2203.03872, 2022: 1-12.
[14]	李磊, 张静, 欧阳齐铖, 等. 基于GRU-VAE 的无监督航迹异常检测方法[J/OL]. 指挥控制与仿真. http://kns.cnki.net/kcms/detail/32.1759.TJ.20221109.1816.008.html.
[15]	KOBYZEV I, PRINCE S J D, BRUBAKER M A. Normalizing flows: an introduction and review of current methods[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 43(11): 3964-3979.
[16]	HUANG C W, KRUEGER D, LACOSTE A, et al. Neural autoregressive flows[C]//The 35th International Conference on Machine Learning, July 10-15, Stockholm, Sweden. New York: PMLR, 2018: 2078-2087.
[17]	DINH L, SOHL-DICKSTIEN J, BENGIO S. Density estimation using real NVP[EB/OL]. (2017-02-27)[2022-08-31]. https://doi.org/10.48550/arXiv.1605.08803.
[18]	KINGMA D P, DHARIWAL P. Glow: generative flow with invertible 1 ×1 convolutions[EB/OL]. (2018-07-10)[2022-8-31]. https://arxiv.org/abs/1807.03039.
[19]	BEHRMANN J, GRATHWOHL W, CHEN R, et al. Invertible residualnetworks[C]//Proceedings of the 36th International Conference on Machine Learning. New York: PMLR, 2019: 573-582.
[20]	NACHMAN B, SHIH D. Anomaly detection with density estimation[J].Physical Review D, 2020, 101(7): 075042.
[21]	WELLHAUSEN L, RANFTL R, HUTTER M. Safe robot navigation via multi-modal anomaly detection[J]. IEEE Robotics and Automation Letters,2020, 5(2): 1326-1333.
[22]	GUDOVSKIY D, ISHIZAKA S, KOZUKA K. CFLOW-AD: real-time unsupervised anomaly detection with localization via conditional normalizing flows[C]//2022 IEEE/CVF Winter Conference on Applications of Computer Vision(WACV), January 3-8, 2022, Waikoloa, HI, USA.IEEE, 2022: 1819-1828.
[23]	ZHOU C, PAFFENROTH R C. Anomaly detection with robust deep autoencoders[C]//Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, August 13-17,2017, Halifax, NS, Canada: ACM, 2017: 665-674.
[24]	RIBEIRO M, LAZZARETTI A E , LOPES H S . A study of deep convolutional auto-encoders for anomaly detection in videos[J]. Pattern Recognition Letters, 2018, 105: 13-22.
[25]	VINCENT P, LAROCHELLE H, LAJOIE I, et al. Stacked denoising autoencoders:learning useful representations in a deep network with a local denoising criterion[J]. Journal of Machine Learning Research, 2010,11(12): 3371-3408.
[26]	CHAUHAN S, VIG L. Anomaly detection in ECG time signals via deep long short-term memory networks[C]//2015 IEEE International Conference on Data Science and Advanced Analytics (DSAA), October 19-21,2015, Paris, France. IEEE, 2015: 1-7.
[27]	CHEN J S, LI J, CHEN W G, et al. Anomaly detection for wind turbines based on the reconstruction of condition parameters using stacked denoising autoencoders[J]. Renewable Energy, 2020, 147: 1469-1480.
[28]	MAKHZANI A, SHLENS J, JAITLY N, et al. Adversarial autoencoders[EB/OL]. (2016-05-25) [2022-08-31]. https://doi.org/10.48550/arXiv.1511.05644.
[29]	CHEN X, KONUKOGLU E. Unsupervised detection of lesions in brain MRI using constrained adversarial auto-encoders[C]//The 1st Medical Imaging for Deep Learning (MIDL 2018), Amsterdam, Netherlands,2018: 04972.
[30]	BEGGEL L, PFEIFFER M, BISCHL B. Robust anomaly detection in images using adversarial autoencoders[C]//ProECML PKDD 2019: Machine Learning and Knowledge Discovery in Databases, September 16-20, 2019, Wurzbury, Germany. Cham: Springer, 2019: 206-222.
[31]	ZHANG H B, GUO W P, ZHANG S Q, et al. Unsupervised deep anomaly detection for medical images using an improved adversarial autoencoder[J]. Journal of Digital Imaging, 2022, 35(2): 153-161.
[32]	PRINCIPI E, VESPERINI F, SQUARTINI S, et al. Acoustic novelty detection with adversarial autoencoders[C]//2017 International Joint Conference on Neural Networks (IJCNN), May 14-19, 2017, Anchorage,AK, USA. IEEE, 2017: 3324-3330.
[33]	CHALAPATHY R, MENON A K, CHAWLA S. Anomaly detection using one-class neural networks[C]//The 24th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, August 19-23, 2018, London,United Kingdom. New York: ACM, 2018: 06360.
[34]	OZA P, PATEL V M. One-class convolutional neural network[J]. IEEE Signal Processing Letters, 2019, 26(2): 277-281.
[35]	MAHBUB U, SARKAR S, PATEL V M, et al. Active user authentication for smartphones: a challenge data set and benchmark results [C]//2016 IEEE 8th International Conference on Biometrics Theory, Applications and Systems (BTAS), September 6-9, 2016, Niagara Falls, NY,USA. IEEE, 2016: 1-8.
[36]	SALEH B, FARHADI A, ELGAMMAL A. Object-centric anomaly detection by attribute-based reasoning[C]//2013 IEEE Conference on Computer Vision and Pattern Recognition, June 23-28, 2013, Portland,OR, USA. IEEE, 2013: 787-794.
[37]	RUFF L, VANDERMEULEN R A, G魻RNITZ N, et al. Deep one-class classification[C]//Proceedings of the 35th International Conference on Machine Learning. New York: PMLR, 2018: 4393-4402.
[38]	VARTOUNI A M, SHOKRI M, TESHNEHLAB M. Auto-threshold deep SVDD for anomaly-based web application firewall[J]. Journal of Latex Class Files, 2015, 14(8): 15135468.
[39]	王晓慧,王延江,邓晓刚, 等. 基于加权深度支持向量数据描述的工业过程故障检测[J].化工学报, 2021, 72(11): 5707-5716.
[40]	CARON S, HENDRIKS L, VERHEYEN R. Rare and different: anomaly scores from a combination of likelihood and out-of-distribution models to detect new physics at the LHC[EB/OL]. (2021-11-12)[2022-08-31].https://arxiv.org/abs/2106.10164.
[41]	ERFANI M, SHOELEH F, GHORBANI A A. Financial fraud detection using deep support vector data description[C]//2020 IEEE International Conference on Big Data (Big Data), December 10-13, 2020, Atlanta, GA, USA. IEEE, 2020: 2274-2282.
[42]	CHEN X Q, CAO C J, MAI J B. Network anomaly detection based on deep support vector data description[C]//2020 5th IEEE International Conference on Big Data Analytics (ICBDA), May 8-11, 2020, Xiamen,China. IEEE, 2020: 251-255.
[43]	RUFF L, VANDERMEULEN R A, G魻RNITZ N, et al. Deep semi-supervised anomaly detection[EB/OL]. (2020-02-14)[2022-08-31]. https://doi.org/10.48550/arXiv. 1906.02694.
[44]	PANG G S, SHEN C H, VAN DEN HENGEL A. Deep anomaly detection with deviation networks[C]//The 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, August 4 - 8, 2019,Anchorage, AK, USA． ACM, 2019: 353-362.
[45]	LU J W , WANG J H, WEI X J, et al. Deep anomaly detection based on variational deviation network[J]. Future Internet, 2022, 14(3): 80.