Joint training method and device for business models

Abstract

The embodiments of this specification provide a method and device for joint training of business models. In multi-party security calculations, nonlinear complex operations are assigned to third parties for processing, which greatly reduces the complexity of joint training of nonlinear models by multiple business parties. At the same time, each business party holding the characteristic data or label data of the training samples generates a predetermined number of random numbers through the agreed random number generation method, so as to obtain the sample data of the training samples corresponding to each other in order to maintain data consistency , and because the third party does not participate in the random number generation process, each other business party effectively guarantees data privacy against the third party. And each other business parties interact through the secret sharing method to ensure data privacy between each other. In short, the above method greatly reduces the computational complexity of joint training of nonlinear models by multiple business parties on the basis of privacy protection.

Description

technical field [0001] One or more embodiments of this specification relate to the field of computer technology, and in particular, to a method and device for training a business model by using a computer. Background technique [0002] In the era of big data, a lot of data may be scattered and stored in different enterprises. Based on privacy protection and other considerations, enterprises cannot fully share each other's data, so data islands may exist. [0003] In some cases, it is usually necessary to carry out cooperative security modeling between enterprises, so as to fully protect the privacy of enterprise data, use the data of all parties to carry out collaborative training on data processing models, so as to process business more accurately and effectively data. At this time, the data used for training the data processing model may be scattered among the parties involved in the cooperative modeling. Therefore, how to provide a more effective model training method b...

AI Technical Summary

Problems solved by technology

[0002] In the era of big data, a lot of data may be scattered and stored in different enterprises. Based on privacy protection and other considerations, enterprises cannot fully share each other's data, so there may be data islands

Images