Parallel selection method of interactive features for large-scale high-dimensional sequence data

Abstract

The invention provides an interactive feature parallel selection method for large-scale high-dimensional sequence data, including: encoding the original high-dimensional SNP data; retaining the encoded SNP data related to the target class through block filtering based on graph theory; Perform fine-grained feature filtering on the SNP data related to the target class; divide the feature-filtered sequence data set into several blocks with γ as the granularity, and obtain the feature candidate area based on the maximum allele common subsequence MACS; corresponding to the candidate area The data set is based on MapReduce for diversity selection of feature regions to obtain representative feature regions; for representative feature regions, the parallel ant colony algorithm of permutation search is used for interactive feature selection to obtain a subset of significant features, that is, significant SNP sites gather. The present invention proposes a brand-new framework for solving interactive feature selection in large-scale sequence data, making feature selection more efficient and more powerful.

Description

technical field [0001] The invention belongs to the technical field of interactive feature selection, in particular to an interactive feature parallel selection method for large-scale high-dimensional sequence data. Background technique [0002] With the continuous advancement and development of data mining and machine learning technology, feature selection technology has received more and more attention. In terms of learning efficiency and learning results, machine learning models have significantly benefited from using only relevant data. The most widely used technique for finding relevant data is feature selection, which is to select a subset of features from the original feature set. The successful application of feature selection also brings new challenges, one of which is to find out the subset of potentially interacting features, because the combination of these features is the subset of features that really affect the target variable (class label). Therefore, the re...

AI Technical Summary

Problems solved by technology

The advantage is that it can be applied to large-scale data sets and is not affected by marginal effects, but its performance is unstable, and the final result depends on the initial value and the data structure of the entire search space

Images