Federated model parameter determination method, device, equipment and storage medium

Abstract

The invention discloses a federated model parameter determination method, a device, equipment and a storage medium. The method comprises the following steps: acquiring a plurality of parameters of a target model, randomly coding and combining the plurality of parameters to obtain a plurality of groups of first parameter codes, respectively executing a federated learning process on the target modelto obtain a plurality of federated models, and determining the accuracy of the plurality of federated models; if the federation models of which the accuracy meets the preset condition in the plurality of federation models are not converged, based on the accuracy of the plurality of federation models, selecting a plurality of groups of second parameter codes from the plurality of groups of first parameter codes to carry out crossover mutation processing to obtain a plurality of groups of third parameter codes, and then executing a federation learning process on the target model; and if the federation models of which the accuracy meets a preset condition in the plurality of federation models are converged, determining a first target parameter coding group corresponding to the converged federation models, and determining a first target parameter corresponding to the first target parameter coding group as a federation model parameter. Therefore, the prediction accuracy of the federated model for different scenes is improved.

Description

technical field [0001] The present invention relates to the technical field of financial technology (Fintech), in particular to a method, device, device and storage medium for determining parameters of a federated model. Background technique [0002] With the continuous development of financial technology (Fintech), especially Internet technology finance, more and more technologies (such as artificial intelligence, big data analysis, cloud storage, etc.) Higher requirements, such as more accurate parameters for generating federated models, are required to improve the accuracy of federated model predictions. [0003] The parameters of the current federated model are usually obtained by training data collected by various sensors, but the built-in parameters of different sensors are different, resulting in different batches of data used for federated model training; the model parameters of the federated model obtained after training Finally, the adaptability of the prediction ...

AI Technical Summary

Problems solved by technology

[0004] The main purpose of the present invention is to provide a federated model parameter determination method, device, device and storage medium, aiming to solve the problem of federated model adaptation caused by model parameters generated by each batch of training data with differences in the prior art. Poor technical issues

Images