Complex mode classification-oriented feature selection method

Abstract

The invention discloses a complex mode classification-oriented feature selection method in the technical field of mode classification. The complex mode classification-oriented feature selection method comprises the following specific steps of S1: discretizing continuous features in a data set D, and representing a result with D; S2: calculating Gini coefficients of the features, and performing repeated operation according to the calculated features; S3: performing ascending order arrangement for values of the features; and S4: finding a rapidly changed point or an inflection point i0 in a broken line graph of the feature set defined in the specification. According to the method, the influence of a clustering threshold on the result is reduced by adopting multi-time repeated calculation; the feature dimension is reduced; the correlation and redundancy after feature extraction are avoided; the working difficulty of a classifier is lowered; and the method can be widely applied to various complex mode classification problems and has the advantages of high self-adaptation capability and wide application range.

Description

technical field [0001] The invention relates to the technical field of pattern classification, in particular to a feature selection method for complex pattern classification. Background technique [0002] Pattern classification problems are currently widely used in various fields of society, such as image classification, data mining, information retrieval, information extraction, speech recognition, etc. The processing methods usually include the following aspects: sample preprocessing, feature extraction, feature selection, classification . Among them, feature selection is an important preprocessing process in pattern classification. In pattern classification, samples after feature extraction often have a large number of features. Feature selection is to filter out features that are irrelevant or less useful for classification from these large number of features, and select features that are very useful for classification, so that the classifier Therefore, feature selecti...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a feature selection method for complex pattern classification, to solve the problem that the feature dimension proposed in the above-mentioned background technology is large, and there may be large correlation and redundancy, which hinders the further extraction of features. and the implementation of the classifier poses a very difficult problem

Images