TCNGANbased Temporal Traffic Anomaly Detection

Abstract

Abstract: In recent years, generative adversarial networks have been widely used in the field of temporal anomaly detection. However, temporal data often has complex timedependence, and problems such as gradient vanishing and training instability are common in existing anomaly detection models. To this end, this paper proposes an unsupervised temporal traffic anomaly detection model based on the combination of temporal convolutional network (TCN) and GAN. The model uses TCN as the infrastructure of generator and discriminator, which can effectively capture the temporal features of the temporal traffic data. During the anomaly detection process, the model constructs an anomaly scoring function based on the reconstruction loss and discriminator loss, and performs anomaly judgment by setting a threshold, thus improving the accuracy of anomaly detection. To verify the performance of the proposed model, experiments are conducted on five different types of datasets. The results show that the average F1 score of the proposed model is 11.02% higher than that of the TAnoGAN model.

Key words: TCN, GAN, unsupervised, anomaly detection, temporal traffic

摘要： 近年来，生成对抗网络在时间序列异常检测领域得到了广泛应用.然而，时序数据往往具有复杂的时间依赖性，而现有异常检测方法中普遍存在梯度消失与训练不稳定等问题.为此，提出了一种基于时序卷积网络(temporal convolutional network, TCN)与生成对抗网络(generative adversarial network, GAN)相结合的无监督时序流量异常检测方法.该方法将TCN作为生成器和辨别器的基础架构，能够有效捕捉时序流量数据的时间特征.异常检测过程中，模型基于重构损失和判别损失构建异常评分函数，并通过设定阈值进行异常判断，从而提高了异常检测的准确性.为验证该模型的性能，在5类不同数据集上进行了实验.结果表明，该模型相比TAnoGAN模型的平均F1分数提高了11.02%.

关键词: 时序卷积网络, 生成对抗网络, 无监督, 异常检测, 时序流量

CLC Number:

TP309

李琛, 林维, 许力, . 基于TCNGAN的时序流量异常检测[J]. 信息安全研究, 2025, 11(10): 907-.

References

［1］Chandola V, Banerjee A, Kumar V. Anomaly detection: A survey［J］.  ACM Computing Surveys, 2009, 41(3): 158［2］Pocher N, Zichichi M,Merizzi F, et al. Detecting anomalous cryptocurrency transactions: An AMLCFT application of machine learningbased forensics［J］.  Electronic Markets, 2023, 33(1): 37［3］Fordal J M, Schjlberg P, Helgetun H, et al. Application of sensor data based predictive maintenance and artificial neural networks to enable Industry 4.0［J］.  Advances in Manufacturing, 2023, 11(2): 248263［4］Li H, Li Y. Anomaly detection methods based on GAN: A survey［J］.  Applied Intelligence, 2023, 53(7): 82098231［5］Choi K, Yi J, Park C, et al. Deep learning for anomaly detection in timeseries data: Review, analysis, and guidelines［J］.  IEEE Access, 2021, 9: 120043120065［6］Goodfellow I, PougetAbadie J, Mirza M, et al. Generative adversarial networks［J］.  Communications of the ACM, 2020, 63(11): 139144［7］Bai S, Kolter J Z, Koltun V. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling［J］.  arXiv preprint, arXiv:1803.01271, 2018［8］Kieu T, Yang B, Jensen C S. Outlier detection for multidimensional time series using deep neural networks［C］ Proc of the 19th IEEE Int Conf on Mobile Data Management (MDM). Piscataway, NJ: IEEE, 2018: 125134［9］陈磊, 秦凯, 郝矿荣. 基于集成LSTMAE的时间序列异常检测方法［J］.  华中科技大学学报: 自然科学版, 2021, 49(11): 3540［10］He Y, Zhao J. Temporal convolutional networks for anomaly detection in time series［C］ Journal of Physics: Conf Series. IOP Publishing, 2019, 1213(4): 042050［11］Nguyen H D, Tran K P, Thomassey S, et al. Forecasting and anomaly detection approaches using LSTM and LSTM autoencoder techniques with the applications in supply chain management［J］.  International Journal of Information Management, 2021, 57: 102282［12］Pena E H M, de Assis M V O, Proena M L. Anomaly detection using forecasting methodsarima and hwds［C］ Proc of the 32nd Int Conf on Chilean Computer Science Society (SCCC). Piscataway, NJ: IEEE, 2013: 6366［13］Jin M, Koh H Y, Wen Q, et al. A survey on graph neural networks for time series: Forecasting, classification, imputation, and anomaly detection［J］.  IEEE Trans on Pattern Analysis and Machine Intelligence, 2024, 46(12): 1046610485［14］Ma J, Perkins S. Timeseries novelty detection using oneclass support vector machines［C］ Proc of the Int Joint Conf on Neural Networks. Piscataway, NJ: IEEE, 2003: 17411745［15］Teng M. Anomaly detection on time series［C］ Proc of the 2010 IEEE Int Conf on Progress in Informatics and Computing. Piscataway, NJ: IEEE, 2010: 603608［16］Ahmed M, Mahmood A N, Hu J. A survey of network anomaly detection techniques［J］.  Journal of Network and Computer Applications, 2016, 60: 1931［17］Cover T, Hart P. Nearest neighbor pattern classification［J］.  IEEE Trans on Information Theory, 1967, 13(1): 2127［18］Breunig M M, Kriegel H P, Ng R T, et al. LOF: Identifying densitybased local outliers［C］ Proc of the 2000 ACM SIGMOD Int Conf on Management of Data. New York: ACM, 2000: 93104［19］MacQueen J. Some methods for classification and analysis of multivariate observations［C］ Proc of the 5th Berkeley Symp on Mathematical Statistics and Probability. Berkeley, CA: University of California Press, 1967: 281297［20］Pang G, Shen C, Cao L, et al. Deep learning for anomaly detection: A review［J］.  ACM Computing Surveys, 2021, 54(2): 138［21］Munir M, Siddiqui S A, Dengel A, et al. DeepAnT: A deep learning approach for unsupervised anomaly detection in time series［J］.  IEEE Access, 2018, 7: 19912005［22］Su Y, Zhao Y, Niu C, et al. Robust anomaly detection for multivariate time series through stochastic recurrent neural network［C］ Proc of the 25th ACM SIGKDD Int Conf on Knowledge Discovery & Data Mining. New York: ACM, 2019: 28282837［23］Provotar O I, Linder Y M, Veres M M. Unsupervised anomaly detection in time series using LSTMbased autoencoders［C］ Proc of the 2019 IEEE Int Conf on Advanced Trends in Information Theory (ATIT). Piscataway, NJ: IEEE, 2019: 513517［24］Kieu T, Yang B, Guo C, et al. Outlier detection for time series with recurrent autoencoder ensembles［COL］ Proc of the 28th Int Joint Conf on Artificial Intelligence. 2019［20250922］. https:vbn.aau.dkenpublicationsoutlierdetectionfortimeserieswithrecurrentautoencoderense［25］Schlegl T, Seebck P, Waldstein S M, et al. Unsupervised anomaly detection with generative adversarial networks to guide marker discovery［C］ Proc of Int Conf on Information Processing in Medical Imaging. Berlin: Springer, 2017: 146157［26］Zenati H, Foo C S, Lecouat B, et al. Efficient GANbased anomaly detection［J］.  arXiv preprint, arXiv:1802.06222, 2018［27］Li D, Chen D, Jin B, et al. MADGAN: Multivariate anomaly detection for time series data with generative adversarial networks［C］ Proc of Int Conf on Artificial Neural Networks. Berlin: Springer, 2019: 703716［28］Bashar M A, Nayak R. TAnoGAN: Time series anomaly detection with generative adversarial networks［C］ Proc of the 2020 IEEE Symp Series on Computational Intelligence (SSCI). Piscataway, NJ: IEEE, 2020: 17781785［29］Geiger A, Liu D, Alnegheimish S, et al. TadGAN: Time series anomaly detection using generative adversarial networks［C］ Proc of the 2020 IEEE Int Conf on Big Data (Big Data). Piscataway, NJ: IEEE, 2020: 3343［30］田腾, 石茂林, 宋学官, 等. 基于滑动窗口的时间序列异常检测方法［J］.  仪表技术与传感器, 2021, 7: 112116［31］Lavin A, Ahmad S. Evaluating realtime anomaly detection algorithms—The Numenta anomaly benchmark［C］ Proc of the 14th IEEE Int Conf on Machine Learning and Applications (ICMLA). Piscataway, NJ: IEEE, 2015: 3844［32］Hundman K, Constantinou V, Laporte C, et al. Detecting spacecraft anomalies using LSTMs and nonparametric dynamic thresholding［C］ Proc of the 24th ACM SIGKDD Int Conf on Knowledge Discovery & Data Mining. New York: ACM, 2018: 387395［33］Hsieh R J, Chou J, Ho C H. Unsupervised online anomaly detection on multivariate sensing time series data for smart manufacturing［C］ Proc of the 12th IEEE Conf on ServiceOriented Computing and Applications (SOCA). Piscataway, NJ: IEEE, 2019: 9097［34］王德文, 潘晓飞, 赵红博. 基于改进生成对抗网络的时序数据异常检测［J］.  计算机工程与设计, 2024, 45(3): 762768

[1]	. Research on Network Unknown Attack Detection Based on Machine Learning#br# #br# [J]. Journal of Information Security Reserach, 2025, 11(9): 807-.
[2]	. A Blackbox Antiforensics Method of GANgenerated Faces Based on #br# Invertible Neural Network#br# [J]. Journal of Information Security Reserach, 2025, 11(5): 394-.
[3]	. A Lightweight Image Steganalysis Model Based on Multidirectional Hybrid Filters [J]. Journal of Information Security Reserach, 2025, 11(4): 318-.
[4]	. Highutility Time Series Data Generation Method Combining Sequence #br# Correlation Graph and GAN#br# [J]. Journal of Information Security Reserach, 2025, 11(4): 351-.
[5]	. An Image Steganography Method Based on Threechannel Deep Fusion Technology [J]. Journal of Information Security Reserach, 2025, 11(3): 257-.
[6]	. Research on Traffic Anomaly Detection Method and System for API Gateway [J]. Journal of Information Security Reserach, 2025, 11(10): 917-.
[7]	. Container Anomaly Detection Based on Attention Mechanism and Multiscale Convolutional Neural Network [J]. Journal of Information Security Reserach, 2025, 11(1): 35-.
[8]	. Malicious Client Detection and Defense Method for Federated Learning [J]. Journal of Information Security Reserach, 2024, 10(2): 163-.
[9]	. Traffic Anomaly Detection Method by Secondorder Feature [J]. Journal of Information Security Reserach, 2024, 10(12): 1082-.
[10]	. Traffic Anomaly Detection Based on Improved Pigeon Inspired Optimizer and #br# Pyramid Convolution#br# [J]. Journal of Information Security Reserach, 2024, 10(12): 1107-.
[11]	. Application of Behavior Anomaly Detection in Zero Trust Access #br# Control Method#br# [J]. Journal of Information Security Reserach, 2024, 10(10): 921-.
[12]	. Cyberbullying Detection Model Based on ELMoTextCNN [J]. Journal of Information Security Reserach, 2023, 9(9): 868-.
[13]	. Research on Blockchain Abnormal Transaction Detection Technology Based on LightGBM [J]. Journal of Information Security Reserach, 2023, 9(9): 877-.
[14]	. Research on Image Steganography and Extraction Scheme Based on Implicit Symmetric Generative Adversarial Network [J]. Journal of Information Security Reserach, 2023, 9(6): 566-.
[15]	. Image Steganalysis Method Based on Multiattention Mechanism and Siamese Network [J]. Journal of Information Security Reserach, 2023, 9(6): 573-.