Gene recognition system of self-adaptation filter based on variable step size minimum mean square error

Abstract

The invention discloses a gene recognition system of a self-adaptation filter based on a variable step size minimum mean square error. The system comprises an improvement unit of a variable step size LMS self-adaptation filter algorithm and a processing unit of a gene structure multi-characteristic weight fusion algorithm, wherein the improvement unit of the variable step size LMS self-adaptation filter algorithm is used to conduct filtering processing of a gene sequence by the variable step size LMS self-adaptation improvement algorithm, so gene characteristics in that random noise is rare, and cycle 3 behaviors are strong can be obtained; and the processing algorithm of the gene structure multi-characteristic weight fusion algorithm is used to extract characteristics of the gene sequence through a multi-characteristic weight fusion strategy, so that characteristic vectors with a higher expression ability can be obtained. The invention proposes the variable step size LMS self-adaptation filter improvement algorithm and the multi-characteristic weight fusion algorithm; and the two algorithms are integrated into the same gene recognition system, so recognition performance of the system is further improved.

Description

technical field [0001] The invention relates to the field of identifying and locating codable gene regions of DNA sequences obtained by sequencing in bioinformatics, and in particular relates to a gene identification system based on a variable step size minimum mean square error (LMS) adaptive filter. Background technique [0002] At present, the research on gene identification has achieved many excellent results, and many mature prediction systems have been formed. However, these systems mainly rely on some classic machine learning labeling methods, such as HMM and CRF. The general model structure is complex and there are many adjustment parameters. , the training is time-consuming and difficult to store, and the system is too specific, which is not conducive to the transfer between species. In addition, it cannot effectively suppress and solve problems such as large random noise in short gene sequences, sparse feature information and low recognition rate. [0003] In orde...

AI Technical Summary

Problems solved by technology

[0002] At present, the research on gene identification has achieved many excellent results, and many mature prediction systems have been formed. However, these systems mainly rely on some classic machine learning labeling methods, such as HMM and CRF. The general model structure is complex and there are many adjustment parameters. , the training is time-consuming and difficult to store, and the system is too specific, which is not conducive to the transfer between species

In addition, it is impossible to effectively suppress and solve problems such as large random noise in short gene sequences, sparse feature information and low recognition rate.

Images