Data Processing Method Based on Incremental Convex Local Nonnegative Matrix Factorization

Abstract

The invention discloses a data processing method based on incremental locally convex non-negative matrix factorization, and mainly aims to solve the problems that negative value-containing data cannot be processed via basic non-negative matrix factorization, and during incremental data processing, the computational consumption is too high and the occupied memory space is relatively large. The data processing method comprises the following steps: 1, expressing initial data as an initial matrix; 2, conducting factorization on the initial matrix according to a locally convex non-negative matrix factorization method, so as to obtain an initial basis matrix and an initial coefficient matrix; 3, receiving new data, and constructing a new data matrix; and 4, conducting factorization on the new data matrix via the initial basis matrix and the initial coefficient matrix according to an incremental locally convex non-negative matrix factorization method, so as to process incremental data. The data processing method can be used for processing the negative value-containing data, so that the application range of basic non-negative matrix factorization is broadened; the physical meaning of the factorization result is definite; the orthogonality, the openness and the efficiency are high; and the data processing method can be used for processing incremental data of images, videos and network access records.

Description

technical field [0001] The invention belongs to the technical field of data processing, relates to a method for processing incremental data, and can be used for data processing of images, videos and network access records. Background technique [0002] With the rapid development of related technologies such as the Internet of Things, sensors, and computers, data is dynamically growing at an explosive rate, such as surveillance video, network access, etc., and the data dimension is getting higher and higher. The task of large-scale high-dimensional data. How to quickly and accurately obtain useful or most needed information from the vast sea of ​​data has become an imminent problem. Data dimensionality reduction is an effective method to solve the "curse of dimensionality". Traditional dimensionality reduction methods include: principal component analysis, independent component analysis, Fisher discriminant analysis, etc. These methods allow negative results, but negative e...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the deficiencies in the prior art above, the present invention proposes a data processing method based on incremental convex local non-negative matrix decomposition to solve the defect that the basic non-negative matrix decomposition method cannot handle negative value data, and at the same time solve the problem of conventional methods in processing incremental data. The problem of low efficiency and serious consumption of computing resources and storage resources when quantifying data

Images