Multi-core clustering method for rapidly processing missing heterogeneous data

Abstract

The invention discloses a multi-core clustering method for rapidly processing the missing heterogeneous data. The multi-core clustering method comprises the following steps of 1, carrying out 0 filling initialization on the missing multi-source heterogeneous data; 2, carrying out multi-core learning on the initialized multi-source heterogeneous data by utilizing a plurality of base kernel functions to generate a multi-core matrix; 3, performing multi-core clustering on the generated multi-core matrix to generate a pseudo label; then carrying out the missing value updating on each basic core matrix forming the multi-core matrix by using low-rank estimation; and 4, based on a clustering result, learning the multi-core joint coefficient by using an extreme learning machine. According to the method, the multi-core clustering method is used for achieving the fast learning of the heterogeneous data, the core completion technology is used for fully completing the information of missing data,the clustering performance is improved, and the problem that a traditional multi-core clustering method cannot effectively process the multi-source heterogeneous data is solved.

Description

technical field [0001] The invention belongs to the field of data mining and machine learning, relates to a multi-core clustering method, in particular to a multi-core clustering method for processing missing heterogeneous data, which can be applied to Web data analysis, biological information analysis, financial investment analysis, Intelligent medical analysis and other fields. Background technique [0002] With the development of the computer field, the concept of "Internet +" has penetrated into all walks of life. In the era of big data, data in these fields has different data formats and sources, and tends to be multi-source heterogeneous. Data from multiple data sources with different types, structures, and distributions is called "multi-source heterogeneous data". For example: the data analyzed when designing a recommendation system may simultaneously contain different types of data such as text, images, and videos from multiple social platforms such as Twitter, Fac...

AI Technical Summary

Problems solved by technology

However, this direct and blunt filling method does not guarantee that the data distribution after filling missing values ​​is consistent with the original data. Therefore, it is easy to reduce the effective information extracted from the data, so a better filling method is needed to improve multi-core clustering. performance

Images