Distance-preserving and dimension-reducing method for data containing missing data

Abstract

The invention discloses a distance-preserving and dimension-reducing method for data containing missing data, and relates to the technical field of data processing. According to the data distance-preserving and dimension-reducing method, the missing part in the original data does not participate in the calculation of the loss function of the automatic encoder through the missing data matrix, so that the automatic encoder can process the dimensionality reduction containing the missing data, and the influence of the missing data on the automatic encoder is avoided; meanwhile, the strong automatic learning capability of the automatic encoder is used; the complex nonlinear relationship between the original data can be effectively captured; and the weight matrix of the encoder in the loss function is updated through constraint, so that the dimension reduction processing has distance preserving performance, the distribution information of the original high-dimensional data is reserved for the low-dimensional data after dimension reduction to the maximum extent, subsequent data processing is facilitated, and the data processing time and space are saved.

Description

technical field [0001] The invention belongs to the technical field of data processing, in particular to an autoencoder-based distance-preserving dimensionality reduction method for data containing missing data. Background technique [0002] With the advent of the era of big data and the popularization of electronic equipment, a large amount of high-dimensional data has been generated. Direct analysis and processing of high-dimensional data usually requires a large time and space overhead. Dimensionality reduction is a method of mapping high-dimensional data to low-dimensional Algorithms that retain the original data information in the dimensional space are increasingly favored by people. Applying dimensionality reduction algorithms such as Principal Components Analysis (PCA for short) or Linear Discriminant Analysis (LDA for short) to reduce the dimensionality of high-dimensional data can bring great convenience to subsequent data processing. However, most of the data gene...

AI Technical Summary

Problems solved by technology

[0011] Aiming at the problems in the prior art that the traditional dimensionality reduction algorithm cannot reduce the dimensionality of missing data, cannot effectively learn the complex nonlinear structural information of high-dimensional data, and does not have distance preservation in the dimensionality reduction process, the present invention provides a method containing The data distance-preserving dimensionality reduction method for missing data achieves dimensionality reduction with missing data by improving the autoencoder, and captures the nonlinear information between high-dimensional data, and then intervenes in the weight matrix in the improved autoencoder , so that the update of the encoder weight matrix conforms to the random projection property, so as to obtain a trained autoencoder, and use the trained autoencoder to reduce the dimensionality of high-dimensional data, which can achieve distance-preserving dimensionality reduction with missing data, so that Save time and space for subsequent data processing

Images