Research on Open Innovation Intelligent Decision-Making of Cross-Border E-Commerce Based on Federated Learning

Table 1

Federated learning averaging algorithm.

Federated averaging algorithm
Input:	Randomly select M(n = 1,…,M)participants from N participants; I is the time of cycle;
	is The initial gradient provided for the server;
	E is the number of time periods on the local SGD (stochastic gradient descent);
	Server: To initialize
	From  = 1，…，T：
	The server sends to ：
	Each client ：
	←Local Update（n， ）
	←
	Local Update（n， ）：
	Number of local device sessions for each participant 1 to E:
	Update according to the learning rate and
	Feedback to the server
Output:	Optimized global gradient