Federated learning system and method based on block chain

Abstract

The invention provides a federated learning system and method based on a block chain, and the system comprises a model training module which is used for the updating of a machine learning model in a federated learning process and the aggregation of the change values of the machine learning model, an intelligent contract module which is based on the blockchain technology and is used for providing adecentralized control function and a key management function in the federated learning process, and a storage module which is based on an IPFS protocol and is used for providing a decentralized information storage mechanism for the intermediate information in the federated learning process, wherein the model training module, the intelligent contract module based on the blockchain technology and the storage module based on the IPFS protocol run on each node participating in federated learning at the same time. Complete decentralization of the whole system is realized, the failure and quit of any node do not influence the continuous federated learning of other nodes, and the robustness is stronger.

Description

technical field [0001] The invention relates to the technical field of federated learning, in particular to a blockchain-based federated learning system and method. Background technique [0002] Federated learning enables participating institutions to collaboratively train machine learning models without directly exchanging raw data. For those enterprises or institutions with insufficient data, it can allow them to join forces to obtain better models without exposing the original data, so as to achieve mutual benefit and win-win results. In the existing engineering technology, the collaborative training of various institutions relies on the centralized third-party collaborative nodes to realize control, aggregation and key management. Existing centralized methods have the following disadvantages: [0003] 1) Collaboration nodes will continuously obtain information uploaded by all other institutions. And a curious collaborative node can use this information to infer import...

AI Technical Summary

Problems solved by technology

[0004] 2) When the cooperative node fails, the entire system will crash and cannot continue to run

Due to the single point of failure of the centralized collaboration node, the federated learning of each institution will be forcibly terminated and the collaborative training cannot continue

[0005] The federated learning method in the prior art has the problems of privacy risk and single point of failure

Images