Active Adaptation for Novelty Detection in Data Streams with Extreme Latency Using Fuzzy Micro-Clusters

• Tiago Carlos Batista UFSCar
• Heloisa Camargo UFSCar

DOI: https://doi.org/10.5753/eniac.2026.31551

Resumo

Data Streams are characterized by continuous data generation, evolving distributions, and limited access to labeled data, posing significant challenges for classification and novelty detection. This paper proposes a fuzzy micro cluster-based approach for these tasks using an active adaptation mechanism that triggers model updates based on recent data batches defined by variable-sized windows, improving adaptability to dynamic environments. A partial update strategy, which uses only a fraction of the accumulated data, is also proposed to reduce the number of model updates while maintaining classification performance. Experiments on synthetic datasets show that the proposed approach achieves similar or superior performance in classification and novelty detection compared to its passive adaptation counterpart, while requiring fewer model updates, indicating reduced computational overhead.

Referências

Abdullatif, A., Masulli, F., and Rovetta, S. (2018). Clustering of nonstationary data streams: A survey of fuzzy partitional methods. WIREs Data Mining and Knowledge Discovery, 8(4):e1258.

Bezdek, J. (1981). Pattern Recognition with Fuzzy Objective Function Algorithms. Plenum Press, New York.

Bruzzone, L. R. D. (2023). Eifuzzcnd: uma estratégia incremental para classificação multiclasse e detecção de novidades em fluxos de dados. Masters thesis, Federal University of São Carlos.

Cristiani, A. and Camargo, H. (2021). A fuzzy multi-class novelty detector for data streams under intermediate latency. In 2021 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE), pages 1–6.

Din, S., Shao, J., Kumar, J., Mawuli, C., Mahmud, S., Zhang, W., and Yang, Q. (2021). Data stream classification with novel class detection: a review. Knowledge and Information Systems, 63:2231–2276.

Ditzler, G., Roveri, M., Alippi, C., and Polikar, R. (2015). Learning in nonstationary environments: A survey. IEEE Computational Intelligence Magazine, 10(4):12–25.

Fahy, C., Yang, S., and Gongora, M. (2022). Scarcity of labels in non-stationary data streams: A survey. ACM Computing Surveys, 55(2):40.

Ferdaus, M., Dam, T., Alam, S., and Pham, D. (2024). X-Fuzz: An evolving and interpretable neuro-fuzzy learner for data streams. IEEE Transactions on Artificial Intelligence, 5(8):4001–4012.

Gama, J., Rodrigues, P., Spinosa, E., and Carvalho, A. (2010). Knowledge discovery from data streams. In Web Intelligence and Security, pages 125–138. Springer.

Gu, X. (2023). A dual-model semi-supervised self-organizing fuzzy inference system for data stream classification. Applied Soft Computing, 136:110053.

Pal, N., Pal, K., Keller, J., and Bezdek, J. (2005). A possibilistic fuzzy c-means clustering algorithm. IEEE Transactions on Fuzzy Systems, 13(4):517–530.

Pimentel, M., Clifton, D., Clifton, L., and Tarassenko, L. (2014). A review of novelty detection. Signal Processing, 99:215–249.

Samadi, N., Tanha, J., and Jalili, M. (2025). A weighted semi-supervised possibilistic fuzzy c-means algorithm for data stream classification and emerging class detection. Knowledge-Based Systems, 309:112831.

Urio-Larrea, A., Camargo, H., Lucca, G., Asmus, T., Marco-Detchart, C., Schick, L., Lopez-Molina, C., Andreu-Perez, J., Bustince, H., and Dimuro, G. (2025). Data stream clustering: introducing recursively extendable aggregation functions for incremental cluster fusion processes. IEEE Transactions on Cybernetics, 55(3):1421–1435.

Wen, K. and Zhao, T. (2026). Interpretable and robust online self-organizing granule-based fuzzy rule classification for data stream. International Journal of Fuzzy Systems, 28:382–396.