Subsection subspace model-based sparse feature extraction and classification method

Abstract

The present invention relates to a feature extraction and classification method for high-dimensional data, in particular to a subsection type subspace model-based sparse feature extraction and classification method. The method includes the following steps that: training data to be processed and test data are preprocessed separately, a spatial identification function is constructed, and a subsection characterization model is constructed; singular value decomposition is performed on the training data of each subsection, and feature dimensionalities corresponding to each subsection are estimated; sparse feature extraction is performed on the training data of the subsections through adopting a subspace learning method and on the basis of the estimated feature dimensionalities; and classification learning is performed on the test data in a feature space through using an elastic network method and on the basis of a sparse feature extraction result, so that a final classification result is obtained. With the subsection type subspace model-based sparse feature extraction and classification method of the invention adopted, the accuracy of the feature dimensionality estimation and feature extraction of local sub-regions can be effectively improved with a relatively small number of samples, and the efficiency and accuracy of existing high-dimensional data classification learning can be effectively improved.

Description

technical field [0001] The invention relates to a sparse feature extraction and classification method for high-dimensional data, in particular to a sparse feature extraction and classification method based on a segmented subspace model. Background technique [0002] With the continuous development of data acquisition technology, a large amount of high-dimensional data that needs to be analyzed and processed has been generated in different application fields, such as medical image data, biological gene data, radar image data, remote sensing image data, voice data, financial data, etc. In recent years, machine learning methods have been widely used in the analysis and learning of big data. High data dimensionality and relatively small number of samples are important problems to be solved in machine learning. The invention mainly relates to a feature extraction and classification method for high-dimensional data. [0003] For high-dimensional data, feature extraction methods ...

AI Technical Summary

Problems solved by technology

But if the training data is not linearly separable, the classification effect of the EN method is poor

Images