This paper studies the remote order data sharing method of self-pick-up machine based on federal learning,accurately and effectively shares the order data of each port operation of self-pick-up machine,and provides a basis for effec-tively analyzing the marketing differences of each port and ensuring its reasonable operation.The self-coding neural network was used to improve the basic federated learning model,and the semi-supervised federated learning model was obtained.Af-ter the model was trained with incremental weighting,the semi-supervised federated learning model was used to share the re-mote order data of each assisted cargo plane port.The results show that the method can effectively share the order data of each remote self-pick-up machine port,and can effectively analyze the best-selling categories of each port in different periods according to the shared data.When there are unmarked data ports and the number of communication wheels is low,the sha-ring accuracy of the method is slightly affected.When the number of communication wheels reaches a certain number,the sharing accuracy of the method is stable without interference from this factor.When the newly added order data exists in the sharing,the newly added order data has almost no effect on the sharing accuracy of the method.
关键词
联邦学习/自助取货机/远程下单数据/自编码/半监督/增量加权
Key words
federated learning/self-service pickup machine/remote order data/self-coding/semi-supervision/incremen-tal weighting
维普
万方数据