Multicomponent model for sequential aggregated assessment of short sequences for system state change detection

S.V. Popereshnyak

Abstract


This paper proposes a multicomponent model for sequential aggregated assessment of short sequences de signed to detect system state changes under a limited number of observations. The relevance of the study is driven by the need for rapid analysis of streaming data in cases where the use of long time series, the accu mulation of large volumes of statistical data, or the training of complex models is impractical or infeasible. The proposed model is based on exact joint distributions of structural statistics of short binary sequences and comprises two interrelated levels of aggregation. At the first level, a local structural-probabilistic assessment of an individual short fragment is formed by considering a set of interrelated characteristics of its internal structure. At the second level, sequential aggregation of an ordered trajectory of local assessments is per formed while preserving information about the nature of their changes. The resulting multicomponent assess ment incorporates characteristics of the level, direction of change, accumulation, duration, and variability of structural atypicality. This approach makes it possible to characterize not only the presence of a local deviation but also the specific features of its formation and evolution throughout the sequence of observations. An experimental study of the model was conducted and confirmed its ability to distinguish between short se quences containing the same number of individual events but exhibiting different internal structures, as well as to differentiate isolated local deviations from persistent and directional structural changes. The model was practically validated using IoT monitoring data from a freshwater ecosystem. The obtained results demon strated the ability to identify different phases of changes in the state of the monitored process and confirmed the practical applicability of the proposed model to streaming monitoring, early detection of structural changes, and the generation of interpretable features for subsequent classification and forecasting tasks.

Problems in programming 2026; 3: 69-82


Keywords


short sequences; binary sequences; structural-probabilistic assessment; sequential aggregation; multicomponent model; system state change detection; structural atypicality; IoT monitoring; streaming data; forecasting

References


Choi, K., Yi, J., Park, C., & Yoon, S. (2021). Deep learning for anomaly detection in timeseries data: Review, analysis, and guidelines. IEEE Access, 9, pp. 120043–120065.

Lindemann, B., Maschler, B., Sahlab, N., & Weyrich, M. (2021). A survey on anomaly detection for technical systems using LSTM networks. Computers in Industry, 131, 103498.

Xu, J., Wu, H., Wang, J., & Long, M. (2022). Anomaly Transformer: Time series anomaly detection with association discrepancy. In International Conference on Learning Representations (ICLR 2022). Retrieved from OpenReview.

Tuli, S., Casale, G., & Jennings, N. R. (2022). TranAD: Deep transformer networks for anomaly detection in multivariate time series data. Proceedings of the VLDB Endowment, 15(6), pp. 1201–1214.

Zhang, C., Zhou, T., Wen, Q., & Sun, L. (2022). TFAD: A decomposition time series anomaly detection architecture with time frequency analysis. In Proceedings of the 31st ACM International Conference on Information & Knowledge Management, pp. 2497–2507. ACM.

Li, G., Yang, Z., Wan, H., & Li, M. (2022). Anomaly-PTG: A time series data-anomaly detection transformer framework in multiple scenarios. Electronics, 11(23), 3955.

Wu, H., Hu, T., Liu, Y., Zhou, H., Wang, J., & Long, M. (2023). TimesNet: Temporal 2D variation modeling for general time series analysis. In International Conference on Learning Representations (ICLR 2023).

Yang, Y., Zhang, C., Zhou, T., Wen, Q., & Sun, L. (2023). DCdetector: Dual attention contrastive representation learning for time series anomaly detection. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pp. 3033–3045. ACM.

Zamanzadeh Darban, Z., Webb, G. I., Pan, S., Aggarwal, C. C., & Salehi, M. (2024). Deep learning for time series anomaly detection: A survey. ACM Computing Surveys, 57(1), Article 15.

Sánchez-Ferrera, A., Calvo, B., & Lozano, J. A. (2025). A review on self-supervised learning in time series anomaly detection: Recent advances and open challenges. ACM Computing Surveys, 58(5), Article 119.

Masol, V. I., & Popereshnyak, S. V. (2020). Checking the randomness of bits disposition in local segments of the (0, 1)-sequence. Cybernetics and Systems Analysis, 56(3), pp. 513–520.

Masol, V. I., & Popereshnyak, S. V. (2021). Joint distribution of some statistics of random bit sequences. Cybernetics and Systems Analysis, 57(1), pp. 160–167.


Refbacks

  • There are currently no refbacks.