Convolutional Neural Networks with 3D Input for P300 Identification in Auditory Brain-Computer Interfaces

Table 3

Proposed convolution and pool patches for the CNN models in which a 2D input was implemented. CP and PP stand for convolution patch and pool patch, respectively.

CP	PP	Accuracy
		500 ms	400 ms	300 ms
		0.781	0.768	0.734
		0.753	0.716	0.727
		0.766	0.725	0.732
		0.724	0.707	0.698
Average		0.756	0.729	0.722