ЎЗ-ЎЗИНИ ТАШКИЛ ЭТУВЧИ КОХОНЕН ХАРИТАЛАРИНИНГ НАЗАРИЙ АСОСЛАРИ ВА АМАЛИЁТДАГИ ҚЎЛЛАНИЛИШИ
Authors
Keywords: Kohonen map, neural network, clustering, unsupervised learning, visualization, data analysis, dimensionality reduction
Abstract
This article presents the theoretical foundations and the learning algorithm of self-organizing Kohonen maps through mathematical models. Kohonen maps are considered an effective neural network model for unsupervised clustering and visualization of data. The paper provides a detailed description of the method's architecture, the process of finding the Best Matching Unit (BMU), the neighborhood model, and the adaptive weight updating process. Additionally, visualization techniques – the Unified Distance Matrix (U-Matrix) and component projections – are discussed. Practical applications of the method in medicine, geospatial data, and agriculture are analyzed, highlighting its importance as a tool for understanding data structures and supporting decision-making.
References
1. Kohonen T. Self-organizing maps (3rd ed.). Springer Berlin, Heidelberg. 2001. –P. 502.
2. Cottrell M., Fort J.C., G. Pagès, Theoretical aspects of the SOM algorithm, Neurocomputing, Volume 21, Issues 1–3, 1998, -P.119-138.
3. Oja E., S. Kaski Kohonen Maps. Amsterdam: Elsevier, 1999. –P. 390.
4. Haykin S.S., Neural Networks: A Comprehensive Foundation 2Nd Ed., Prentice-Hall Of India Pvt. Limited. 1999. – P. 842.