Model joint training method and device based on block chain

Abstract

The embodiment of this specification provides a block chain-based model joint training method and device. In the training method, the current layer node of the block chain receives the first training result sent by the corresponding upper layer node. The first training result is at least determined based on the first type of result. The first type of result is obtained by the upper layer node after training the target model among the various models maintained on the chain based on the private data maintained off the chain. Based on the private data maintained off the chain, the current layer node trains the target model among the various models maintained on the chain, and obtains the second training result. The first training result and the second training result are fused to obtain a fusion result. Determine whether the current layer node is at the last layer. If the current layer node is at the last level, broadcast the fusion result on the block chain, so that multiple nodes update the target model among the various models maintained on the chain based on the fusion result.

Description

technical field [0001] One or more embodiments of this specification relate to the field of computer technology, and in particular to a blockchain-based model joint training method and device. Background technique [0002] Federated Learning, also known as federated learning, is an emerging basic technology of artificial intelligence. Its design goal is to carry out efficient machine learning among multiple participants (or computing nodes or subjects) under the premise of ensuring information security during big data exchange, protecting terminal data and personal data privacy, and ensuring legal compliance. [0003] In the traditional model joint learning or training process, in order to ensure the security of the data of all parties, the data exchanged between the parties is usually encrypted, but there is still a risk of data leakage. [0004] Therefore, it is necessary to provide a more secure method for joint training of models. Contents of the invention [0005] O...

AI Technical Summary

Problems solved by technology

[0003] In the traditional model joint learning or training process, in order to ensure the security of the data of all parties, the data exchanged between the parties is usually encrypted, but there is still a risk of data leakage

Images