Federation learning model training method, device and system and storage medium

Abstract

The invention discloses a federation learning model training method, device and system and a storage medium. The method comprises the following steps of sending a model updating request message carrying the joint model parameters and the waiting duration to each participating device; receiving the model parameter update which is sent when each participating device carries out local training on the to-be-trained model according to the local data and the joint model parameters and determines that time participates in the model update; updating the model parameters to obtain the latest joint model parameters; counting the participation state of each participation device for updating the model, and obtaining the waiting duration for updating the model next time according to the counting result, wherein the model updating request message of the next model updating carries the latest joint model parameters and the waiting duration of the next model updating until when the to-be-trained model is in a convergence state, the latest joint model parameters are used as the final parameters of the to-be-trained model. According to the method, the training time and quality of the federation learning model are well considered simultaneously.

Description

technical field [0001] The present invention relates to the technical field of data processing, in particular to a federated learning model training method, device, system and storage medium. Background technique [0002] With the development of artificial intelligence, in order to solve the problem of data islands, people put forward the concept of "federated learning", so that both sides of the federation can also conduct model training to obtain model parameters without giving their own data, and can avoid data The issue of privacy breaches. [0003] In the current model parameter update process of horizontal federated learning, each participating device only uses its own locally owned data to train the model, and sends model parameter updates to the coordinating device, and the coordinating device updates the model parameters received from different participating devices Fusion is carried out, and the updated model parameters after fusion are distributed to each partici...

AI Technical Summary

Problems solved by technology

[0005] The main purpose of the present invention is to provide a federated learning model training method, equipment, system and storage medium, aiming to solve the problem that the training time and model quality of the federated learning model cannot be well considered in the current horizontal federated learning model training process

Images