Data clustering method based on bivariant weighted kernel FCM algorithm

Abstract

The invention discloses a data clustering method based on a bivariant weighted kernel FCM algorithm. The data clustering method comprises the steps that firstly, a data set is divided optimally to minimize an objective function; an initial membership degree matrix, a typical value matrix and an initial clustering center are obtained; the distances between data points in a multi-kernel high-dimensional space and the clustering center are obtained; iteration is conducted to obtain membership degree values and probability typical values; and a clustering result corresponding to the minimum valueobtained by the objective function is adopted as a final clustering result. According to the data clustering method, a kernel function guided by a combined kernel is adopted to replace an ordinary Euclidean distance function, and thus linear data and nonlinear data can be better divided; and the anti-noise property of the algorithm is improved through the typical value matrix, the accuracy of algorithm clustering is improved, the proportion of various kernels in the combined kernel can be automatically adjusted to meet the requirements of different datasets for different kernel functions, andthe problem of selection uncertainty of an ordinary kernel algorithm to the kernel functions is solved.

Description

technical field [0001] The invention relates to the technical field of data clustering, in particular to a data clustering method based on a bivariate weighted kernel FCM algorithm. Background technique [0002] Clustering is an important research content in the field of data mining and artificial intelligence, and it has played a great role in many fields such as big data, pattern recognition, image segmentation, and machine learning. Clustering is the process of dividing data according to the similarity rules of the data. The result of the division is determined by the rules. The divided groups or sets are often called clusters. Fuzzy c-means clustering algorithm (FCM) is the most basic method of fuzzy clustering, which lays the foundation for clustering algorithms based on objective functions, but this method is not only sensitive to the initialization of cluster centers, but also susceptible to noise points. In order to improve the noise immunity of the algorithm, Krish...

AI Technical Summary

Problems solved by technology

[0006] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a data clustering method based on the bivariate weighted kernel FCM algorithm, in order to accurately avoid the fact that FCM is sensitive to noise points, and PCM is easy to produce consistent clustering problems, increase the accuracy of the algorithm, and more accurately mine the data structure information existing in the data set

Images