Cluster analysis of data with the use of artificial neural networks

DOI №______

Abstract

The article describes the process of cluster analysis of data - the analysis of data, which is based on the association of objects that have common properties in the group. The task of clustering is relevant, since the growing accumulation of a large number of data leads to the need to classify them in the light of an increasing number of parameters, so the task of developing and applying methods that specialize in the classification of multidimensional data sets. The article considers and analyzes the contribution of scientists to the study of cluster analysis of data using artificial neural networks and their achievement in the coming years, the feasibility of using artificial neural networks for the process of cluster analysis of data. The model of the system of cluster analysis of data using artificial neural networks is described with the use of mathematical apparatus. For a model of cluster analysis of data based on neural networks, wellknown structures of k-medium clustering with differences in KL were considered on the basis of soft pairwise constraints. The proposed k-mean bounded algorithm uses some marked data to control the uncontrolled k-medium clustering. Unlike the random initialization of cluster centers in traditional k-averages, marking samples are used to initiate cluster centers in restricted k-averages. Also, with each iteration, the k-mean reassignment of the cluster is limited to unmarked samples, and the membership of the labeled samples is fixed. This procedure of limited k-averages showed performance improvements by the k-medium algorithm. The described method makes it possible to automate the process of cluster analysis of data, especially when the number of clusters from the beginning is unknown. To do this, based on known k-medium methods and KL differences, a model of cluster analysis system based on the neural network was described.

Keywords: cluster; clustering; cluster analysis; algorithm; mathematical model; artificial neural network; big data.

References
1. Deep clustering via joint convolutional autoencoder embedding and relative entropy minimization / K. G. Dizaji, A. Herandi, C. Deng [et al.] // In 2017 IEEE International Conference on Computer Vision (ICCV). IEEE, 2017. P. 5747–5756.
2. Rich feature hierarchies for accurate object detection and semantic segmentation / R. Girshick, J. Donahue, T. Darrell, J. Malik // In CVPR, 2014. P. 580–587.
3. Improved deep embedded clustering with local structure preservation / X. Guo, L. Gao, X. Liu, J. Yin // In International Joint Conference on Artificial Intelligence (IJCAI-17), 2017. P. 1753–1759,
4. Deep clustering with convolutional autoencoders / X. Guo, X. Liu, E. Zhu, J. Yin // In International Conference on Neural Information Processing, Springer, 2017. P. 373–382.
5. Xie J., Girshick R., Farhadi A. Unsupervised deep embedding for clustering analysis // In International conference on machine learning, 2016. P. 478–487.
6. Yang J., Parikh D., Batra D. Joint unsupervised learning of deep representations and image clusters // In CVPR, 2016. P. 5147–5156.