A data privacy protection method for federated learning

Abstract

The invention discloses a data privacy protection method for federated learning, which includes two steps of autonomy and association. Specifically, the autonomy is: two or more participants install initialized models on their respective terminals, and each participant Having the same model, the participants each use local data to train the model to obtain different model parameters; the joint is specifically: the participants upload the different model parameters to the cloud at the same time, and the cloud completes the aggregation and integration of the model parameters. update, and return the updated parameters to the terminals of each participant, and the terminals of each participant start the next iteration, and repeat the above steps until the entire training process converges. The invention realizes the joint modeling under the condition that the data does not go out of the local area, and uses the interaction of the model parameters instead of the direct exchange of the data, which not only realizes the data interaction, but also solves the privacy and security problems of the data.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a data privacy protection method of federated learning. Background technique [0002] Artificial intelligence is extremely dependent on scenarios and data. High-quality data directly determines the quality of the model, but data acquisition has become difficult. This dilemma mainly lies in two aspects: on the one hand, the phenomenon of isolated data islands is serious, and most of the data is held in the three "data island groups" of the government, operators, and Internet companies, and the utilization rate of data is low and the cost is high; on the other hand, On the one hand, data security and user privacy issues are getting more and more attention, data supervision is becoming more stringent, and data interoperability is becoming more difficult. Contents of the invention [0003] In order to solve the above problems, the present invention provides a data privacy protec...

AI Technical Summary

Problems solved by technology

[0002]Artificial intelligence is extremely dependent on scenarios and data. High-quality data directly determines the quality of the model, but data acquisition has become difficult

This dilemma mainly lies in two aspects: on the one hand, the phenomenon of data islands is serious, and most of the data is held in the three "data island groups" of the government, operators, and Internet companies, and the utilization rate of data is low and the cost is high; on the other hand On the one hand, data security and user privacy issues are getting more and more attention, data supervision is becoming more stringent, and data interoperability is becoming more difficult

Images