Quantum image-based feature selection method

Abstract

The invention provides a Quantum image-based feature selection method, and the method comprises the following steps: obtaining d feature values of the quantum image, and obtaining an n-row d-column quantum data set D; combining d quantum features of the quantum data set D into a set FS; selecting a subset F in the set FS, and dividing the quantum data set D into M sets; extracting all d one-dimensional feature sets circularly from the set FS, calculating contribution degrees, and reserving the feature set with the maximum contribution degree; extracting all d * (d-1) two-dimensional feature sets circularly from the set FS, calculating contribution degrees, and reserving the feature set with the maximum contribution degree; sequentially and circularly extracting three-dimensional, four-dimensional,... and d-dimensional feature sets, calculating contribution degrees, retaining the feature set with the maximum contribution degree, and circularly comparing to obtain a finally selected feature set. According to the invention, how to calculate the contribution degree of a quantum image feature set to quantum machine learning by using a quantum entropy method is given on a quantum computer, then quantum image feature selection is carried out, and the method plays a promoting role in research of quantum machine learning and popularization of application of the quantum computer.

Description

technical field [0001] The invention relates to the field of machine learning, in particular to a feature selection method based on quantum images. Background technique [0002] Machine learning is to obtain a model by analyzing the attributes or characteristics of sample objects. when classifying images. Images generally have many features, some of which are critical to building a model, and some are useless. Therefore, before machine learning, it is necessary to select image features that are useful for building a model, that is, image feature selection, that is, to select all image feature sets. the best subset of . In order to select the optimal subset, it is necessary to calculate each quantum image feature set, that is, to determine which quantum image feature set is the optimal feature set. On a quantum computer, due to the unobservable and non-copyable nature of the quantum state, the traditional feature set evaluation algorithm is invalid. Therefore, it is necess...

AI Technical Summary

Problems solved by technology

On a quantum computer, due to the unobservable and non-copyable nature of the quantum state, the traditional feature set evaluation algorithm is invalid. Therefore, it is necessary to establish an evaluation method for the quantum image feature set for the quantum computing environment, that is, how to calculate the quantum image feature set pair The contribution of quantum machine learning, and then feature selection

Images