Attentive 3D-Ghost Module for Dynamic Hand Gesture Recognition with Positive Knowledge Transfer

Algorithm 1

Training process of the whole framework

	Input: dataset ; : the first training epochs ( is set to 60);  E: the total number of training epochs  (E is set to 75 in this paper);
	Output: parameter sets and for 3DGSAI network of modality m and modality n;
	1: Training starts:
	2: Initialize and
	3: fordo
	4:
	5:  update and with loss function and through
	6: end for
	7: fordo
	8:  extract the output features and with 3DGSAI network of modality m and n
	9:  Obtain corr and corr by formula corr  =
	10:  Calculate with equation (9)
	11: ifthen
	12:
	13: else
	14:
	15: end if
	16:  update and with loss function and through
	17: end for
	18: Training finishes