Hybrid Deep-Learning Framework Based on Gaussian Fusion of Multiple Spatiotemporal Networks for Walking Gait Phase Recognition

Algorithm 1

Training process of the Gaussian fusion layer based on EM-Algorithm.

	Initialize and set estimation parameter
For T iterations do
Procedure M-step:
Calculate the expectations of (7):
Calculate partial derivatives of :
Return,
Procedure E-step:
Calculate i-th submodel responsivity:
Return
Update the estimation parameter:
End For till the parameter threshold:
or
End Procedure M-step and E-step
Returnand the fusion score S