Table 1 Construction of the 1D-ResCNN designed.

From: A hybrid optimization-enhanced 1D-ResCNN framework for epileptic spike detection in scalp EEG signals

Layer (type)

Output Shape

InputLayer

(None, 20, 1)

Conv1D (layer-1)

(None, 20, 64)

Swish-1

(None, 20, 64)

BatchNormalization-1

(None, 20, 64)

Conv1D (layer-2)

(None, 20, 64)

Swish-2

(None, 20, 64)

BatchNormalization-2

(None, 20, 64)

Residual connection-1

(None, 20, 64)

Conv1D (layer-3)

(None, 20, 64)

Swish-3

(None, 20, 64)

BatchNormalization-3

(None, 20, 64)

Residual connection-2

(None, 20, 64)

Conv1D (layer-4)

(None, 20, 64)

Swish-4

(None, 20, 64)

BatchNormalization-4

(None, 20, 64)

Residual connection-3

(None, 20, 64)

Dropout

(None, 20, 64)

Flatten

(None, 768)

Dense-1

(None, 256)

Dense-2

(None, 512)

Dense-3

(None, 2)

Back to article page