Image dimension reduction method of extreme learning machine based on graph embedding

Abstract

The invention provides an image dimension reduction method of an extreme learning machine based on graph embedding, and belongs to the field of machine learning and data mining. The method comprises the following steps: firstly, selecting an original image sample set, and constructing a sample relation matrix by utilizing inter-sample distances and label information of original image samples; secondly, according to the constructed sample relation matrix, firstly, carrying out random mapping on an input vectorized image sample, and then, learning a feature extraction matrix by minimizing a weighted sample reconstruction error; and finally, performing data dimension reduction on vectorized image data by using the learned feature extraction matrix. The method is short in training time and efficient in data compression, and the compression quality and dimension reduction stability of the data are effectively improved.

Description

technical field [0001] The invention belongs to the field of machine learning and data mining, and in particular relates to an image dimensionality reduction method of an extreme learning machine based on graph embedding. Background technique [0002] With the advent of the era of big data, the data volume and data dimension of today's data are showing a rapid upward trend, and the problem of high data dimension becomes particularly obvious in image data. When performing data analysis and processing, in the face of these high-dimensional data, data compression can effectively avoid the problem of "dimensional explosion" and reduce the storage burden; in addition, data compression can effectively remove redundancy in the original data features to improve post-data processing performance. An important branch of data compression is the dimensionality reduction algorithm. [0003] Traditional dimensionality reduction algorithms such as principal component analysis (PCA), linea...

AI Technical Summary

Problems solved by technology

However, most of today's self-encoders based on extreme learning machines start from the perspective of unsupervised learning, that is, only the data itself is used, and a large amount of valuable label information is ignored.

In addition, the traditional self-encoder based on extreme learning machine uses a random matrix to project the original samples in the early stage of extraction, which may destroy the distance relationship between the original samples, resulting in unstable performance of the self-encoder

Images