Brain Informatics

Table 1 CNN-LSTM structure

From: Channel-independent recreation of artefactual signals in chronically recorded local field potentials using machine learning

Layer	Type	Description
1	sequenceInput	–
2	sequenceFolding	–
3	convolution2d	size=5, filters=32, dilation=1
4	batchNormalization+elu	–
5	convolution2d + elu	size=5, filters=32,dilation=2
6	convolution2d + elu	size=5, filters=32,dilation=4
7	convolution2d + elu	size=5, filters=32,dilation=8
8	convolution2d + elu	size=5, filters=32,dilation=16
9	averagePooling2d	size=1,stride=5
10	sequenceUnfolding	with flattening
11	gru	128
12	lstm	64
13	dropout	0.25
14	lstm	32
15	dropout	0.25
16	regression	–

Back to article page