Convolutional Neural Network for Seizure Detection of Nocturnal Frontal Lobe Epilepsy

Figure 2

CNN architecture, where I is the image input layer; CU_k is the k^th convolutional unit; MP and AP are the max pooling and average pooling layers, respectively; FC is the fully connected layer; and SM and CO are the SoftMax and classification output layer, respectively. In the k^th convolutional unit, C_kj, N_kj, R_kj, and D_k are the j^th convolutional layer, the j^th batch normalization layer, the j^th ReLu layer, and the dropout layer, respectively.