Ensuring Accuracy and Stability of Multidimensional Data Clustering Using Kohonen Self-Organizing Maps Based on Automatic Data Reduction

Read full text Summarise with VIRAT
Authors: Олександр Іващенко, Serhii Fedin
Cited by: 0
References: 20

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The article investigates a method for improving the accuracy and stability of multidimensional data clustering by combining Kohonen self-organizing maps with an automatic feature reduction procedure. A method for removing insignificant features using an interquartile range algorithm integrated into a software application for building SOMs is proposed. A comparison of the clustering results before and after reduction is performed with unchanged training parameters and a fixed seed value, a parameter that determines the initial state of the random number generator and ensures the reproducibility of the initialization of the neuron weight coefficients. It has been proven that the removal of features with low variability does not violate the topological structure of SOM, preserves the stability of cluster separation, and reduces computational costs. A high degree of correspondence between the data before and after reduction has been confirmed based on statistical analysis. It has been established that the proposed method increases the efficiency of SOM and is suitable for analyzing large multidimensional samples

Fields of Study & Subjects

Neural Networks and Applications Data Mining and Machine Learning Applications Advanced Clustering Algorithms Research

Publication details

Journal / source
Artificial Intelligence
Volume
31
Issue
AI.2026.31(1)
Pages
71–89
Publication date
27 March 2026
Language
EN