Cross-Subject Seizure Detection in EEGs Using Deep Transfer Learning

Figure 4

(a) The deep transfer model from VGG16, where the output layer is replaced by a new softmax output layer with two neurons, corresponding to seizure or nonseizure. (b) The deep transfer model from ResNet50, where the output layer is replaced by a new softmax output layer with two neurons, corresponding to seizure or nonseizure, and the size of full connection layers are also altered.

(a)

(b)