Feature selection algorithm for particle swarm hybrid optimization in combination with collaborative learning strategy

Abstract

The invention belongs to the field of machine learning, and particularly relates to a particle swarm hybrid optimization feature selection algorithm combined with a collaborative learning strategy, and the technical scheme is as follows: 1, carrying out coarse-grained feature selection by using a Fisher score and MIC hybrid filtering algorithm, and obtaining a coarse-grained feature subset; 2, performing fine-grained feature selection by using an adaptive particle swarm optimization algorithm of a collaborative learning strategy to obtain an optimal feature subset; according to the method, the problem of high-dimensional data falling into local optimum and high calculation cost in the prior art is effectively solved.

Description

technical field [0001] The invention belongs to the field of machine learning, and in particular relates to a feature selection algorithm for particle swarm hybrid optimization combined with collaborative learning strategies. Background technique [0002] In the medical field, people's demand for data mining is increasing day by day. At the same time, the development of bioinformatics has led to a sharp increase in the dimensions of a large amount of medical data, resulting in high-dimensional data. For example, microarray data sets and gene expression profiles are typical high-dimensional data sets. Analyzing these data, that is, classifying data in the field of machine learning, is helpful for the diagnosis and treatment of diseases. However, in practice, a major challenge in classifying such data is that the data dimension is high and the sample size is small, that is, the problem of "high-dimensional small samples". Classical classification algorithms cannot achieve the ...

AI Technical Summary

Problems solved by technology

[0005] The present invention provides a feature selection algorithm of particle swarm hybrid optimization combined with collaborative learning strategy. The purpose is to solve feature selection by combining the advantages of filtering and packaging feature selection algorithms through the particle swarm hybrid optimization strategy based on collaborative learning strategy. In the process, the calculation cost is high, the subsequent classification performance is low, and the problem of falling into a local optimal solution enables the selected feature subset to obtain better classification ability in a smaller dimension.

Images