Abstract
Automatic high-level feature extraction has become a possibility with the advancement of deep learning, and it has been used to optimize efficiency. Recently, classification methods for Convolutional Neural Network (CNN)-based electroencephalography (EEG) motor imagery have been proposed, and have achieved reasonably high classification accuracy. These approaches, however, use the CNN single convolution scale, whereas the best convolution scale varies from subject to subject. This limits the precision of classification. This paper proposes multibranch CNN models to address this issue by effectively extracting the spatial and temporal features from raw EEG data, where the branches correspond to different filter kernel sizes. The proposed method’s promising performance is demonstrated by experimental results on two public datasets, the BCI Competition IV 2a dataset and the High Gamma Dataset (HGD). The results of the technique show a 9.61% improvement in the classification accuracy of multibranch EEGNet (MBEEGNet) from the fixed one-branch EEGNet model, and 2.95% from the variable EEGNet model. In addition, the multibranch ShallowConvNet (MBShallowConvNet) improved the accuracy of a single-scale network by 6.84%. The proposed models outperformed other state-of-the-art EEG motor imagery classification methods.
Highlights
With the introduction of sophisticated machine learning algorithms, high-performance computers, edge and cloud computing, and next-generation communication technologies, smart healthcare has become a reality [1,2]
This paper focuses on EEG signals based on motor imagery (MI), which means imagining the movement of limbs without moving them
The nature of MI EEG signals gives an advantage to convolutional neural network (CNN) compared to other deep learning architectures
Summary
With the introduction of sophisticated machine learning algorithms, high-performance computers, edge and cloud computing, and next-generation communication technologies, smart healthcare has become a reality [1,2]. The biggest gap between humans and machines is being bridged with the use of brain–computer interfaces (BCIs) Advances in this field enable computers to be deliberately managed by brain signal activity monitoring [3]. They have temporal and spatial features coming from the time spent imagining the movement and, at the same time, the record from different electrodes (each electrode has different locations that contain the spatial information) For this reason, CNNs have several advantages in MI EEG data processing: (i) raw data can be inputted into the system, thereby eliminating the need for prior feature extraction; (ii) the ability to learn temporal and spatial features at the same time; (iii) the ability to exploit the hierarchical nature of certain signals; (iv) high accuracy on large datasets
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
