Anomaly Detection in Streaming Data Using Attention Mechanism-Based Clustered Isolation Forest
DOI:
https://doi.org/10.14419/gcj42498Published
14-06-2026Keywords:
Anomaly Detection; Streaming Data; Isolation Forest; X-Means Clustering; Attention MechanismAbstract
As the need to perform real-time analytics in multiple fields (finance, healthcare, internet of things, etc.) increases, the ac-curacy of anomaly detection in streaming data is becoming a crucial requirement. Conventional techniques are susceptible to high-velocity, non-stationary settings, as they rely on global information and are unable to adjust to concept drift. To address these shortcomings, a new hybrid model, Attention Mechanism-based Clustered Isolation Forest (AM_C_IForest), is introduced here. The method takes advantage of X-means clustering, which performs dynamic data partitioning, the Huberian contamination model, which provides resilience to noise, and an attention mechanism with an optimization prob-lem that uses Nadaraya-Watson kernel regression, which assigns adaptive weights to isolation trees using instance-level relevance. Experimental results across eight benchmark datasets reveal that AM_C_IForest achieves averages of 0.92 & of 0.88 for AUC-ROC and AUC-PR, respectively, outperforming conventional methods such as LOF (AUC-ROC: 0.81, AUC-PR: 0.74), One-Class SVM (AUC-ROC: 0.78, AUC-PR: 0.69), and standard I-Forest (AUC-ROC: 0.85, AUC-PR: 0.79). The results of the study highlight the accuracy, flexibility, and scalability of the model in identifying anomalies that take place in real-time in high-complexity data streaming settings.
References
[1] Bashar, M. A., & Nayak, R. (2020, December). TAnoGAN: Time series anomaly detection with generative adversarial networks. In 2020 IEEE Symposium Series on Computational Intelligence (SSCI) (pp. 1778-1785). IEEE. https://doi.org/10.1109/SSCI47803.2020.9308512.
[2] Blázquez-García, A., Conde, A., Mori, U., & Lozano, J. A. (2021). A review of outlier/anomaly detection in time series data. ACM computing surveys (CSUR), 54(3), 1-33. https://doi.org/10.1145/3444690
[3] Boniol, P., Linardi, M., Roncallo, F., & Palpanas, T. (2020, April). Automated anomaly detection in large sequences. In 2020 IEEE 36th international conference on data engineering (ICDE) (pp. 1834-1837). IEEE. https://doi.org/10.1109/ICDE48307.2020.00182.
[4] Boniol, P., Linardi, M., Roncallo, F., Palpanas, T., Meftah, M., & Remy, E. (2021). Unsupervised and scalable subsequence anomaly detection in large data series. The VLDB Journal, 30(6), 909-931. https://doi.org/10.1007/s00778-021-00655-8.
[5] Braei, M., & Wagner, S. (2020). Anomaly detection in univariate time-series: A survey on the state-of-the-art. arXiv preprint arXiv:2004.00433.
[6] Chalapathy, R., & Chawla, S. (2019). Deep learning for anomaly detection: A survey. arXiv preprint arXiv:1901.03407.
[7] Chen, R. Q., Shi, G. H., Zhao, W. L., & Liang, C. H. (2021). A joint model for IT operation series prediction and anomaly detection. Neurocompu-ting, 448, 130-139. https://doi.org/10.1016/j.neucom.2021.03.062.
[8] Basheer, M. Y. I., Ali, A. M., Hamid, N. H. A., Ariffin, M. A. M., Osman, R., Nordin, S., & Gu, X. (2024). Autonomous anomaly detection for streaming data. Knowledge-Based Systems, 284, 111235. https://doi.org/10.1016/j.knosys.2023.111235
[9] Rajakumar, R., & Devi, S. S. (2024). An efficient modelling of oversampling with optimal deep learning enabled anomaly detection in streaming data. China Communications, 21, 249–260. https://doi.org/10.23919/JCC.ja.2022-0592.
[10] Anomaly Detection in Resource Constrained Environments with Streaming Data. (2022). IEEE Transactions on Emerging Topics in Computational Intelligence, 6(3), 649–659. https://doi.org/10.1109/TETCI.2021.3070660.
[11] Yang, C., Du, Z., Meng, X., Zhang, X., Hao, X., & Bader, D. A. (2023). Anomaly Detection in Catalog Streams. IEEE Transactions on Big Data, 9, 294–311. https://doi.org/10.1109/TBDATA.2022.3161925.
[12] Zhang, M., Guo, J., Guo, J., Li, X., Li, X., Jin, R., & Jin, R. (2020). Data-Driven Anomaly Detection Approach for Time-Series Streaming Data. Sensors, 20(19), 5646. https://doi.org/10.3390/s20195646.
[13] Anomaly Detection in Catalog Streams. (2023). IEEE Transactions on Big Data, 9(1), 294–311. https://doi.org/10.1109/TBDATA.2022.3161925.
[14] Munir, M., Siddiqui, S. A., Chattha, M. A., Dengel, A., & Ahmed, S. (2019). FuseAD: Unsupervised Anomaly Detection in Streaming Sensors Data by Fusing Statistical and Deep Learning Models. Sensors, 19(11), 2451. https://doi.org/10.3390/s19112451.
[15] Le, K.-N. T., Dang, T.-B., Le, D. T., Raza, S. M., Kim, M., & Choo, H. (2023). VEAD: Variance profile Exploitation for Anomaly Detection in re-al-time IoT data streaming. Internet of Things. https://doi.org/10.1016/j.iot.2023.100994.
[16] Cook, A. A., Mısırlı, G., & Fan, Z. (2019). Anomaly detection for IoT time-series data: A survey. IEEE Internet of Things Journal, 7(7), 6481-6494. https://doi.org/10.1109/JIOT.2019.2958185.
[17] Gabriel Rodriguez Garcia, Gabriel Michau, Mélanie Ducoffe, Jayant Sen Gupta, and Olga Fink. 2020. Time Series to Images: Monitoring the Condi-tion of Industrial Assets with Deep Learning Image Processing Algorithms. arXiv: 2005.
[18] E. Keogh, T. Dutta Roy, U. Naik, and A Agrawal. 2021. Multidataset time-series anomaly detection competition. (2021).
[19] Chunggyeom Kim, Jinhyuk Lee, Raehyun Kim, Youngbin Park, and Jaewoo Kang. 2018. DeepNAP: deep neural anomaly pre-detection in a semi-conductor fab. Information Sciences, 457–458, 1–11. doi: 10.1016/j.ins.2018.05.020.
[20] Dan Li, Dacheng Chen, Baihong Jin, Lei Shi, Jonathan Goh, and See-Kiong Ng. 2019. MAD-GAN: Multivariate Anomaly Detection for Time Se-ries Data with Generative Adversarial Networks. In Proceedings of the International Conference on Artificial Neural Networks, 703–716. doi: 10.1007/978- 3- 030-30490-4_56.
[21] Mohsin Munir, Shoaib Ahmed Siddiqui, Andreas Dengel, and Sheraz Ahmed. 2019. DeepAnT: A Deep Learning Approach for Unsupervised Anomaly Detection in Time Series. IEEE Access, 7, 1991–2005. doi: 10.1109/ACCESS.2018. 2886457.
[22] Linardi, M., Zhu, Y., Palpanas, T., & Keogh, E. (2020). Matrix profile goes MAD: variable-length motif and discord discovery in data series. Data Mining and Knowledge Discovery, 34(4), 1022-1071. https://doi.org/10.1007/s10618-020-00685-w.
[23] Maharaj, E. A., D'Urso, P., & Caiado, J. (2019). Time series clustering and classification. Chapman and Hall/CRC. https://doi.org/10.1201/9780429058264.
[24] Marteau, P. F., Soheily-Khah, S., & Béchet, N. (2017). Hybrid isolation forest-application to intrusion detection. arXiv preprint arXiv:1705.03800.
[25] Schmidl, S., Wenig, P., & Papenbrock, T. (2022). Anomaly detection in time series: a comprehensive evaluation. Proceedings of the VLDB Endow-ment, 15(9), 1779-1797. https://doi.org/10.14778/3538598.3538602.