Fast feature selection method based on mixed-feature KDE conditional-entropy matrix formula

Abstract

The present invention provides a fast feature selection method based on the mixed feature KDE conditional entropy matrix, which proposes the concepts of kernel matrix, data vector, partition matrix and kernel partition matrix, and based on these concepts, proposes a mixed feature KDE The entropy matrix greatly improves the calculation efficiency of the mixed feature KDE conditional entropy, thereby improving the efficiency of the feature selection method based on the mixed feature KDE conditional entropy matrix.

Description

Technical field [0001] The present invention relates to a feature selection method, in particular to a fast feature selection method based on a mixed feature KDE conditional entropy matrix formula. Background technique [0002] Feature selection is an important preprocessing process in data mining, machine learning, pattern recognition and other fields. It performs feature selection on multi-dimensional data, but loses redundant and task-related features, which improves the efficiency of learning tasks. [0003] The feature selection method based on information theory is one of the effective methods in the feature selection method, but the traditional feature selection method based on information theory is only for discrete data. The feature selection algorithm based on KDE entropy utilizes the effectiveness of kernel density estimation (KDE) for the parameter-free estimation of continuous random variables, and effectively unifies the continuous attributes and discrete attributes i...

AI Technical Summary

Problems solved by technology

However, the calculation time of KDE entropy is a major drawback of KDE entropy

Images