| Title |
EEG Signal Feature Extraction Algorithm using Takens Embedding |
| Authors |
이기배(Kibae Lee) ; 고건혁(Guhn Hyeok Ko) ; 이종현(Chong Hyun Lee) |
| DOI |
https://doi.org/10.5573/ieie.2023.60.10.55 |
| Keywords |
EEG signal; Takens embedding; Nonlinear feature; CSP |
| Abstract |
Since brain waves are generated by interactions of numerous neurons, interpretation based on linear model has limitations. We propose a nonlinear feature extraction method based on Takens embedding. The method proposed computes the distance between samples of the Electroencephalogram (EEG) signal represented in multidimensional phase space via Takens embedding and obtains the statistical dispersion. We can also improve discrimination by using common spatial pattern (CSP) in the feature extraction procedures. We obtained an improved AUC of 63.59% compared to existing methods using the proposed nonlinear feature in the experiment using publicly available EEG data. In addition, we obtained an improved AUC of 70.98% using the proposed feature in the target detection experiment designed with multiple satellite images compared to using the existing features. Furthermore, we were able to obtain an improved AUC up to 77.81% by applying the CSP. |