A Multichannel Convolutional Neural Network Architecture for the Detection of the State of Mind Using Physiological Signals from Wearable Devices

Table 7

Multichannel CNN architecture.

Layer	Layer type	Filters	Size	No. of parameters	Output dimension	Activation
1	Input	—	—	—	ECG: (4, 1); EMG: (3, 1); RESP: (3, 1); BVP: (3, 1); ACCL: (15, 1)	—
2	Conv1D (1st layer)	128	ECG: (2, 1); EMG: (2, 1); RESP: (2, 1); BVP: (2, 1); ACCL: (8, 1)	ECG: 384; EMG: 384; RESP: 384; BVP: 384; ACCL: 1152	ECG: (3, 128); EMG: (2, 128); RESP: (2, 128); BVP: (2, 128); ACCL: (8, 128)	ReLU
3	Conv1D (2nd layer)	64	ECG: (2, 1); EMG: (2, 1); RESP: (2, 1); BVP: (2, 1); ACCL: (8, 1)	ECG: 16448; EMG: 16448; RESP: 16448; BVP: 16448; ACCL: 65600	ECG: (2, 64); EMG: (1, 64); RESP: (1, 64); BVP: (1, 64); ACCL: (1, 64)	ReLU
4	Flatten		—	—	ECG: 128; EMG: 64; RESP: 64; BVP: 64; ACCL: 64	—
5	Dropout	—	—	—	—
6	Dense (1st layer)	64	—	ECG: 8256; EMG: 4160; RESP: 4160; BVP: 4160; ACCL: 4160	ECG: 64; EMG: 64; RESP: 64; BVP: 64; ACCL: 64	ReLU
7	Dropout	—	—	—	—	—
8	Dense (2nd layer)	32		ECG: 2080; EMG: 2080; RESP: 2080; BVP: 2080; ACCL: 2080	ECG: 32; EMG: 32; RESP: 32; BVP: 32; ACCL: 32	ReLU
9	Concatenate	—		0	160
10	Dense (3rd layer)	32	160	5152	32	ReLU
11	Dense (output)	5	32	165	5	SoftMax