Multiscale Bidirectional Input Convolutional and Deep Neural Network for Human Activity Recognition

Table 1

MultiConv2D & ComplexConv1D network parameters. Each layer of convolution passes through the BN layer; the selected activation function is ReLU.

MultiConv2D		ComplexConv1D
Layer name	Kernel size, filters/stride, pad	Layer name	Kernel size, filters/stride, pad
Conv_1 MaxPooling	, , same	Conv1d_1	, /1, same
Conv_2	,, same	Conv1d_2	, /1, same
Conv_3	, , same	Conv1d_3	, /1, same
Conv_4	, , same	GlobalAveragePooling1D
GlobalAveragePooling2D