TY - GEN
T1 - The Classification of Wink-Based EEG Signals
T2 - Innovative Manufacturing, Mechatronics and Materials Forum, iM3F 2020
AU - Mahendra Kumar, Jothi Letchumy
AU - Rashid, Mamunur
AU - Musa, Rabiu Muazu
AU - Mohd Razman, Mohd Azraai
AU - Sulaiman, Norizam
AU - Jailani, Rozita
AU - P. P. Abdul Majeed, Anwar
N1 - Funding Information:
The present study is funded by Universiti Malaysia Pahang via RDU180321.
Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
PY - 2022
Y1 - 2022
N2 - Electroencephalogram (EEG) is non-trivial in the diagnosis and treatment of neurogenerative diseases. Brain-Computer Interface (BCI) that utilises EEG is often used to improve the activities of daily living of patients with the aforesaid disorder. In this study, the efficacy of different Transfer Learning (TL) models, i.e., ResNet50, ResNet101 and ResNet152 in extracting features to classify wink-based EEG signals is evaluated. The time–frequency spectrum transformation of the Right-Wink, Left-Wink, and No-Wink based on EEG signals was achieved via Discrete Wavelet Transform (DWT). The extracted features were then fed into different variation of Support Vector Machine (SVM) classifiers to evaluate the performance of the different feature extraction method in classifying the wink class. The data are divided into training, validation, ad test, with a stratified ratio of 60:20:20. It was shown from the study, that the features extracted via ResNet152 were better than that of ResNet50 and ResNet101. The overall validation and test accuracy attained through the ResNet152 model is approximately 92%. Henceforth, it could be concluded that the proposed pipeline suitable to be adopted to classify wink-based EEG signals for different BCI applications.
AB - Electroencephalogram (EEG) is non-trivial in the diagnosis and treatment of neurogenerative diseases. Brain-Computer Interface (BCI) that utilises EEG is often used to improve the activities of daily living of patients with the aforesaid disorder. In this study, the efficacy of different Transfer Learning (TL) models, i.e., ResNet50, ResNet101 and ResNet152 in extracting features to classify wink-based EEG signals is evaluated. The time–frequency spectrum transformation of the Right-Wink, Left-Wink, and No-Wink based on EEG signals was achieved via Discrete Wavelet Transform (DWT). The extracted features were then fed into different variation of Support Vector Machine (SVM) classifiers to evaluate the performance of the different feature extraction method in classifying the wink class. The data are divided into training, validation, ad test, with a stratified ratio of 60:20:20. It was shown from the study, that the features extracted via ResNet152 were better than that of ResNet50 and ResNet101. The overall validation and test accuracy attained through the ResNet152 model is approximately 92%. Henceforth, it could be concluded that the proposed pipeline suitable to be adopted to classify wink-based EEG signals for different BCI applications.
KW - BCI
KW - Classification
KW - DWT
KW - EEG
KW - SVM
KW - Transfer learning
UR - http://www.scopus.com/inward/record.url?scp=85112492522&partnerID=8YFLogxK
U2 - 10.1007/978-981-33-4597-3_6
DO - 10.1007/978-981-33-4597-3_6
M3 - Conference Proceeding
AN - SCOPUS:85112492522
SN - 9789813345966
T3 - Lecture Notes in Electrical Engineering
SP - 61
EP - 70
BT - Recent Trends in Mechatronics Towards Industry 4.0 - Selected Articles from iM3F 2020
A2 - Ab. Nasir, Ahmad Fakhri
A2 - Ibrahim, Ahmad Najmuddin
A2 - Ishak, Ismayuzri
A2 - Mat Yahya, Nafrizuan
A2 - Zakaria, Muhammad Aizzat
A2 - P. P. Abdul Majeed, Anwar
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 6 August 2020 through 6 August 2020
ER -