Title: Fast feature selection algorithm of EEG data based on GPU technology

Authors: Bin Liu; Han Deng; Tianke Fang; Meixuan Chen

Addresses: College of Software Engineering, Xiamen University of Technology, Xiamen, Fujian, 361024, China ' Big Data Development and Research Center, Guangzhou College of Technology and Business, Guangzhou, 528138, China ' College of Software Engineering, Xiamen University of Technology, Xiamen, Fujian, 361024, China ' Big Data Development and Research Center, Guangzhou College of Technology and Business, Guangzhou, 528138, China

Abstract: With the rapid development of EEG technology, the rapid selection of EEG data features makes it possible for subsequent applications. To improve the relevance of EEG data, this paper proposes a feature selection algorithm of EEG data based on GPU technology. This method takes the whole gene pathway as a whole, takes its EEG data path as a specific variable, and quantifies each path by distance measurement; secondly, it uses regularisation dimension reduction technology to brush out the essential features of EEG data and makes use of this method to select the essential features of EEG data. Finally, principal component analysis (PCA) is used to evaluate the path model quickly and accurately. The experimental analysis shows that the method solves the problems of high requirement of EEG data sample distribution and slow evaluation speed. Moreover, the method proposed in this paper shows the great ability of real-time performance.

Keywords: EEG; disease characteristics; GPU; distance metric; principal component analysis; PCA.

DOI: 10.1504/IJES.2021.121081

International Journal of Embedded Systems, 2021 Vol.14 No.6, pp.602 - 610

Received: 07 Aug 2020
Accepted: 09 Nov 2020
Published online: 24 Feb 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article