Federal learning method, device and system based on block chain

Abstract

One or more embodiments of the invention disclose a federated learning method, device and system based on a block chain, and the method introduces a block chain technology on the basis of the existing federated learning, and deploys each node participating in the federated learning as a block chain node, therefore, on-chain maintenance management is performed on some important data of federal learning based on the smart contract on the block chain, so that the model training cooperation process is transparent and reliable. Specifically, a block chain node trains model version data corresponding to an updated model according to local training data to perform uplink evidence storage, the model does not need to be uploaded, and the model version data is directly requested to be acquired from a publisher when needed, so that the dependence on centralized services is reduced, and invalid transmission is reduced. Meanwhile, the evaluation result of the on-chain evidence storage model can be verified, and the disguised behaviors of participants can be frightened and prevented.

Description

technical field [0001] This document relates to the field of blockchain technology, and in particular to a blockchain-based federated learning method, device and system. Background technique [0002] Blockchain is a decentralized and innovative solution that uses distributed ledger technology to solve multi-party trust problems, and is a cutting-edge technology in the current society. [0003] Federated learning is a distributed machine learning paradigm that can effectively solve the problem of data islands, allowing participants to jointly model without sharing data, technically breaking data islands, and realizing collaboration in machine learning training models. [0004] At present, when training the federated learning model, due to the participation of multiple parties and the training data provided is not data plaintext, the reliability of the model released by the participating party cannot be guaranteed, and the behavior of maliciously providing wrong data by the pa...

AI Technical Summary

Problems solved by technology

[0004] At present, when training the federated learning model, due to the participation of multiple parties and the training data provided is not data plaintext, the reliability of the model released by the participating party cannot be guaranteed, and the behavior of maliciously providing wrong data by the participating party cannot be avoided, resulting in federated learning. Efficiency is not guaranteed

Images