Federal learning-based block chain data management method and system, and storage medium

Abstract

The invention discloses a block chain data management method and system based on federated learning, and a storage medium, and the method comprises the steps: carrying out the decentralized federated learning based on a block chain, enabling a single terminal device or a plurality of terminal devices to obtain a block output weight through competition, and enabling the terminal device obtaining the block output weight to serve as an aggregator of the current learning; the aggregator aggregates all qualified local model parameters and obtains global model parameters, all parameters of the round are stored in a block, and after the global model is updated and discharged from the block, the terminal equipment obtains the global parameters from the block chain and enters the next round of learning; an incentive mechanism: the terminal equipment participating in federal learning obtains incentive according to the contribution; and equipment selection: carrying out equipment selection based on the game theory and the Markov dynamic decision process. The terminal equipment participating in machine learning is stimulated, and acquisition of a large amount of data is facilitated; a block chain technology is used for tracing in a data calculation link, and hostile attacks are blocked.

Description

technical field [0001] The invention relates to a block chain data management method, system and storage medium based on federated learning. Background technique [0002] There are currently two main data asset management systems on the market. One is government data banks, which are mainly based on blockchain-based government big data asset management and sharing service platforms. The model is based on the management and operation of bank currency assets. Model, with the protection of user data ownership, management rights, use rights and revenue rights as the core, to establish a management and operation system for big data assets, and to construct a third-party intermediary integration of data rights confirmation, aggregation, storage, hosting, sharing and value-added applications The service platform mainly adopts the data service model to carry out data asset-based operation services. It belongs to a big data sharing economy platform. Its main functions include data ag...

AI Technical Summary

Problems solved by technology

[0003] The original data machine learning must transfer data to a centralized server for calculation, and there is still a risk of data leakage; at the same time, federated learning technology is facing more and more malicious attacks, and it is necessary to track tampered data. A malicious data detection method such as Beta reputation calculation, privacy calculation, etc., all need to master the data provided by the user. To ensure the accuracy, this detection method needs to collect a large amount of data for training the detection model

Images