Recognition of motor imagery EEG signals based on capsule network

In order to fully extract the temporal and spatial features contained in motor imagery electroencephalography (EEG) signals for effective identification of imagery, a three-dimensional capsule network (3D-CapsNet) EEG signal recognition model is proposed, which can integrate MI-EEG dimension, channel dimension intrinsic relationship between maximize feature representation capability. Firstly, multi-layer 3D convolution module used time inter-channel space dimensions as low-level features. Secondly, advanced are also obtained through integration. Finally, dynamic routing connections squash functions applied classification. The experimental analysis conducted on BCI competition IV dataset 2a. proposed performs well all subjects’ datasets, such that average accuracy Kappa value 9 subjects 84.028% 0.789, respectively. results confirm effectiveness method. Additionally, four-class classification improved, impact individual variability overcome certain extent.