Why the study?
The study investigated the feasibility of applying deep learning technologies to classify filtered EEG alpha-band signals under a relatively low data volume scenario.
A convolutional neural network provides higher accuracy than a multi-layer perceptron for classifying EEG signals in a brain-computer interface for a wheeled robot, though it requires more computational resources.
Answers are grounded in this paper and cite their sources.
Higher CNN accuracy for EEG-based robotic BCI may aid control; leaves open whether resource demands limit clinical translation.
The article describes the essence of functioning of the neural computer interface, as well as provides the prototype of the custom neural computer control system, which includes the helmet with the Open BCI Cyton platform, BCI-server on the basis of the personal computer and, the wheeled robot itself with the on-board computer Raspberry Pi. Transmission of the recorded 16-channel EEG-records onto the BCI-server is performed using the Bluetooth protocol, and the Wi-Fi standard is applied for the communication between the robot and the BCI-server. The main task was to create and research the possibility of application of the deep learning technologies to classification of the filtered signals (frequency band of the EEG Alpha-waves) under relatively low data volume scenario. Program architecture and system functioning algorithm are presented, convolutional neural network and the multi-layer perceptron are researched as the neural classifiers. EEG-signals filtering and their classification are performed on the BCI-server. Neuroclassifier on the basis of the convolutional neural network showed higher accuracy, however it demands bigger calculating resources for its realization.
No takes yet. Share an insight, caveat, or question.
Artemuk et al. (2021) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: