Improved empirical mode decomposition bagging RCSP combined with Fisher discriminant method for EEG feature extraction and classification

Dongyi Chen ¹

⁰College of Electrical Engineering and Automation Fuzhou University, NO.2, Wulong Jiangbei Avenue, Fuzhou University Town, Minhou, Fuzhou City, Fujian Province, China.

Heliyon +

|

April 1, 2024

View abstract on PubMed

Summary

A new algorithm improves Electroencephalogram (EEG) signal classification accuracy. The improved Empirical Mode Decomposition Bagging Regularized Common Spatial Pattern (EMD Bagging RCSP) method enhances robustness in small datasets.

Area Of Science

Neuroscience
Signal Processing
Machine Learning

Background

Traditional Common Spatial Pattern (CSP) algorithms struggle with noise sensitivity and low accuracy in small Electroencephalogram (EEG) datasets.
This limitation hinders reliable EEG signal classification for various applications.

Purpose Of The Study

To develop an improved algorithm for robust EEG signal classification, particularly for small sample sizes.
To enhance the accuracy and reliability of EEG analysis by addressing noise and data limitations.

Main Methods

An improved Empirical Mode Decomposition (EMD) Bagging Regularized Common Spatial Pattern (RCSP) algorithm was proposed.
The method utilizes improved EMD for noise filtering and feature extraction, Bagging for data reconstruction, and regularization with Fisher linear discriminant analysis for classification.

Main Results

The EMD Bagging RCSP algorithm demonstrated improved accuracy and robustness over traditional CSP methods.
A significant increase of approximately 6% in the average classification rate was observed, validating the algorithm's effectiveness.
The algorithm successfully retained effective information while inhibiting high-frequency noise in small sample EEG datasets.

Conclusions

The proposed EMD Bagging RCSP algorithm offers a reliable and effective solution for EEG signal classification.
This method holds potential for diverse applications, including brain-computer interfaces and clinical EEG diagnostics.

Keywords:

Bagging RCSP Electro encephalon graph (EEG) Feature extraction and classification Fisher discriminant analysis Improved empirical mode decomposition