Skip to main content
. 2021 Oct 22;15:760979. doi: 10.3389/fnins.2021.760979

Figure 2.

Figure 2

Illustration of the convolutional neural network (CNN)-based model. T = number of the timestamps, E = number of electricodes, F1 = number of filters in the temporal CNN, D = number of depthwise convolution output channels, F2 = number of spatial filters, P1 and P2 are sizes of the average pooling kernel, and αlsr and αct are weights of label smoothing regularization and center loss.