Mouse Down's syndrome key protein screening method based on gene-enhanced skeleton particle swarm optimization feature selection algorithm

Abstract

The invention discloses a mouse Down's syndrome key protein screening method based on a gene-enhanced skeleton particle swarm optimization feature selection algorithm, which comprises the following steps: randomly dividing an original data set into two parts, namely a training set and a test set, for original mouse cerebral cortex protein detection, which are used for optimizing a particle swarm and testing a final result; carrying out dimension extraction on the data set to initialize a population to obtain an original population; calculating the fitness of each dimension of gene in the initial population, determining whether the gene of the dimension is selected or not, calculating the accuracy of particle prediction according to the selected gene, and identifying the key category protein for detecting the Down's syndrome of the mouse. In the aspect of particle swarm optimization, compared with a traditional mouse Down's syndrome key protein screening method, the method can quickly and efficiently identify a small number of key feature subsets with good classification performance in an original data set.

Description

technical field [0001] The invention belongs to the application field of computer analysis technology of feature selection, and relates to a method for screening key proteins of mouse Down's syndrome using a gene-enhanced backbone particle swarm optimization feature selection algorithm. Background technique [0002] Pattern recognition can be used in medical diagnosis and remote sensing, speech and text recognition, etc. This method uses computer technology to classify or identify a group of processes or events. The identified data, processes, or events can be specific objects such as sounds, images, and text, or abstract objects such as status and degree. These objects are distinguished from information data in digital form and the like, and are called schema information. In specific applications, this method has achieved many results in X-ray photo analysis, cancer cell detection, chromosome analysis, blood testing, EEG diagnosis and electrocardiogram diagnosis. [0003]...

AI Technical Summary

Problems solved by technology

Although the above methods have improved the convergence or diversity of the algorithm to a certain extent, the performance of these algorithms often decreases when faced with complex problems such as multimodal problems or non-convex problems.

In addition, if no new global optimal solution is generated during the iterative process, and some particles do not generate a new historical optimal solution at the same time, the search range of these particles will be limited within a certain range. In practice, Enhancing the predictive performance of feature subsets is the main purpose of feature selection and there must be a higher probability of falling into local optimum

Images