Clustering Based on Multilayer Self-organizing Kohonen Maps
DOI:
https://doi.org/10.24160/1993-6982-2026-1-140-148Keywords:
clustering, neural network, self-organizing Kohonen mapAbstract
Multi-layer Kohonen’s self-organizing feature maps (MLKSFM) are considered in the clustering problem. The trained MLKSFM predicts the cluster number of the data element received at its input. Algorithms for training and determining the MLKSFM prediction are provided, and examples of data visualization on different model layers are given. The clustering quality is estimated using the ACC indicator, which represents the clustering accuracy. The experiment in which this estimate is determined for MLKSFM and conventional clustering (CC) methods on the Iris and MNIST datasets has shown that on data with a small number of features (Iris, 4 features), the MLKSFM outperforms the CC, and that both the MLKSFM and CC are ineffective on data characterized by a large number of features (MNIST, 784 features). Ways of improving the clustering quality based on the MLKSFM are analyzed.
References
1. Ren Y. e. a. Deep Clustering: a Comprehensive Survey // IEEE Trans. Neural Netw. Learn. Syst. 2025. V. 36(4). Pp. 5858—5878.
2. Kohonen T. The Self-organizing Map // Proc. IEEE. 1990. V. 78(9). Pp. 1464—1480.
3. Popular Unsupervised Clustering Algorithms [Электрон. ресурс] https://scikit-learn.org/stable/api/sklearn.cluster.html (дата обращения 01.05.2025).
4. Lu Shen, Segall R.S. Multi-SOM: an Algorithm for High-Dimensional, Small Size Datasets // Proc. XVI World Multi-conf. Systemics, Cybernetics and Informatics. 2012. V. 1. Pp. 236—241.
5. Nakagawa A., Kutics A. Classification in Big Image Datasets Using Layered-SOM // Proc. XIII Intern. Conf. Signal-image Technol. & Internet-based Systems. 2017. Pp. 143—150.
6. Ichimura T., Yamaguchi T. A Proposal of Interactive Growing Hierarchical SOM // Proc. IEEE Conf. Systems, Man and Cybernetics. Anchorage, 2011. Pp. 3149—3154.
7. Sakkari M., Zaied M. A Convolutional Deep Self-organizing Map Feature Extraction for Machine Learning // Multimedia Tools and Applications. 2020. V. 79. Iss. 27—28. Pp. 19451—19470.
8. Gharaee1 Z. Online Recognition of Unsegmented Actions with Hierarchical SOM Architecture // Cogn. Process. 2021. V. 22. Pp. 77—91.
9. Iris Species [Электрон. ресурс] https://www.kaggle.com/datasets/uciml/iris (дата обращения 01.05.2025).
10. MNIST Database [Электрон. ресурс] https://en.wikipedia.org/wiki/MNIST_database (дата обращения 01.05.2025).
11. Guérin A., Chauvet P., Saubion F. A Survey on Recent Advances in Self-organizing Maps [Электрон. ресурс] https://arxiv.org/abs/2501.08416 (дата обращения 01.05.2025).
12. Хайкин С. Нейронные сети. М.: Издат. дом Вильямс, 2006.
13. Deshmukh A. Variational Quantum Self-organizing Map [Электрон. ресурс] https://arxiv.org/pdf/2504.03584 (дата обращения 01.05.2025).
14. Müller A. Self-organizing Kohonen Map in Python with Periodic Boundary Conditions [Электрон. ресурс] https://github.com/alexarnimueller/som (дата обращения 01.05.2025).
15. Fränti P., Sieranoja S. Clustering Accuracy // Appl. Computing and Intelligence. 2024. V. 4(1). Pp. 24—44.
16. SciPy Documentation [Электрон. ресурс] https://docs.scipy.org/doc/scipy/index.html (дата обращения 01.05.2025).
17. Scikit-learn. Machine Learning in Python. [Электрон. ресурс] https://scikit-learn.org/stable/index.html (дата обращения 01.05.2025).
18. McConville R. N2D: (Not Too) Deep Clustering via Clustering the Local Manifold of an Autoencoded Embedding // Proc. XXV Intern. Conf. Pattern Recognition. 2021. Pp. 5145—5152.
19. Lim K. Deep Clustering Using Adversarial Net-based Clustering Loss. [Электрон. ресурс] https://arxiv.org/html/2412.08933v1 (дата обращения 01.05.2025).
20. Yang X. e. a. Adversarial Learning for Robust Deep Clustering // NeurIPS Proc. 2020. V. 33. Pp. 9098—9108.
21. Wang J., Jiang J. Unsupervised Deep Clustering via Adaptive GMM Modeling and Optimization // Neurocomputing. 2021. V. 433. Pp. 199—211.
22. Yang B. e.a. Towards K-means-friendly Spaces: Simultaneous Deep Learning and Clustering // Proc. XXXIV Intern. Conf. Machine Learning. 2017. V. 70. Pp. 3861—3870.
---
Для цитирования: Бартеньев О.В. Кластеризация на основе многослойных самоорганизующихся карт Кохонена // Вестник МЭИ. 2026. № 1. С. 140—148. DOI: 10.24160/1993-6982-2026-1-140-148
#
1. Ren Y. e. a. Deep Clustering: a Comprehensive Survey. IEEE Trans. Neural Netw. Learn. Syst. 2025;36(4):5858—5878.
2. Kohonen T. The Self-organizing Map. Proc. IEEE. 1990;78(9):1464—1480.
3. Popular Unsupervised Clustering Algorithms [Elektron. Resurs] https://scikit-learn.org/stable/api/sklearn.cluster.html (Data Obrashcheniya 01.05.2025).
4. Lu Shen, Segall R.S. Multi-SOM: an Algorithm for High-Dimensional, Small Size Datasets. Proc. XVI World Multi-conf. Systemics, Cybernetics and Informatics. 2012;1:236—241.
5. Nakagawa A., Kutics A. Classification in Big Image Datasets Using Layered-SOM. Proc. XIII Intern. Conf. Signal-image Technol. & Internet-based Systems. 2017:143—150.
6. Ichimura T., Yamaguchi T. A Proposal of Interactive Growing Hierarchical SOM. Proc. IEEE Conf. Systems, Man and Cybernetics. Anchorage, 2011:3149—3154.
7. Sakkari M., Zaied M. A Convolutional Deep Self-organizing Map Feature Extraction for Machine Learning. Multimedia Tools and Applications. 2020;79;27—28:19451—19470.
8. Gharaee1 Z. Online Recognition of Unsegmented Actions with Hierarchical SOM Architecture. Cogn. Process. 2021;22:77—91.
9. Iris Species [Elektron. Resurs] https://www.kaggle.com/datasets/uciml/iris (Data Obrashcheniya 01.05.2025).
10. MNIST Database [Elektron. Resurs] https://en.wikipedia.org/wiki/MNIST_database (Data Obrashcheniya 01.05.2025).
11. Guérin A., Chauvet P., Saubion F. A Survey on Recent Advances in Self-organizing Maps [Elektron. Resurs] https://arxiv.org/abs/2501.08416 (Data Obrashcheniya 01.05.2025).
12. Khaykin S. Neyronnye Seti. M.: Izdat. Dom Vil'yams, 2006. (in Russian).
13. Deshmukh A. Variational Quantum Self-organizing Map [Elektron. Resurs] https://arxiv.org/pdf/2504.03584 (Data Obrashcheniya 01.05.2025).
14. Müller A. Self-organizing Kohonen Map in Python with Periodic Boundary Conditions [Elektron. Resurs] https://github.com/alexarnimueller/som (Data Obrashcheniya 01.05.2025).
15. Fränti P., Sieranoja S. Clustering Accuracy. Appl. Computing and Intelligence. 2024;4(1):24—44.
16. SciPy Documentation [Elektron. Resurs] https://docs.scipy.org/doc/scipy/index.html (Data Obrashcheniya 01.05.2025).
17. Scikit-learn. Machine Learning in Python. [Elektron. Resurs] https://scikit-learn.org/stable/index.html (Data Obrashcheniya 01.05.2025).
18. McConville R. N2D: (Not Too) Deep Clustering via Clustering the Local Manifold of an Autoencoded Embedding. Proc. XXV Intern. Conf. Pattern Recognition. 2021:5145—5152.
19. Lim K. Deep Clustering Using Adversarial Net-based Clustering Loss. [Elektron. Resurs] https://arxiv.org/html/2412.08933v1 (Data Obrashcheniya 01.05.2025).
20. Yang X. e. a. Adversarial Learning for Robust Deep Clustering. NeurIPS Proc. 2020;33:9098—9108.
21. Wang J., Jiang J. Unsupervised Deep Clustering via Adaptive GMM Modeling and Optimization. Neurocomputing. 2021;433:199—211.
22. Yang B. e.a. Towards K-means-friendly Spaces: Simultaneous Deep Learning and Clustering. Proc. XXXIV Intern. Conf. Machine Learning. 2017;70:3861—3870
---
For citation: Bartenyev O.V. Clustering Based on Multilayer Self-organizing Kohonen Maps. Bulletin of MPEI. 2026;1:140—148. (in Russian). DOI: 10.24160/1993-6982-2026-1-140-148
