Information bottleneck-based possibility fuzzy joint clustering method

Abstract

The invention provides an information bottleneck-based possibility fuzzy joint clustering method. An information bottleneck principle is introduced, mutual information loss of the information bottleneck is adopted as a distance measurement mode, and advantages of possibility clustering and fuzzy joint clustering are combined at the same time to standardize the centroid in the clustering process. After the method is adopted, the sensitivity to initial values can be reduced, the robustness and the anti-noise ability are better, the clustering precision is higher, and a more distinct fuzzy partition result is generated.

Description

technical field [0001] The invention relates to a possible fuzzy joint clustering method based on information bottleneck, which belongs to the field of Internet data processing. Background technique [0002] Recent studies have pointed out that the total number of Internet sites in the world has exceeded 1 billion, and this number is still increasing. People's access to knowledge is gradually shifting from traditional books and newspapers to the Internet. With the rapid development of Internet technology, the amount of data is also increasing. Classifying data is a basic requirement for data processing and analysis. Clustering is an unsupervised data partitioning technique. According to the similarity and dissimilarity between objects, the data is aggregated into several clusters, so that the similarity of the data in the cluster can be maximized, and the dissimilarity of the data between the clusters can be maximized. . Now there are many mature clustering methods, but th...

AI Technical Summary

Problems solved by technology

[0003] The existing fuzzy joint clustering and possibility clustering each have their own advantages and limitations. Fuzzy joint clustering is stable and suitable for high-dimensional sparse matrix clustering, but its performance is greatly affected by outliers; possibility The clustering method can better deal with outliers, but if there is no good initialization, consistent clustering often occurs; and in high-dimensional data, the Euclidean distance cannot measure the similarity between objects very well

Images