A model distributed training method based on blockchain node trust management

By utilizing node-local training behavior data to determine trust values ​​and manage trust during distributed training, the problem of insufficient global model reliability in existing technologies is solved, thereby improving the security of distributed training and the quality of the global model.

CN120012168BActive Publication Date: 2026-07-24HARBIN INSTITUTE OF TECHNOLOGY (SHENZHEN) (INSTITUTE OF SCIENCE AND TECHNOLOGY INNOVATION HARBIN INSTITUTE OF TECHNOLOGY SHENZHEN)
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
HARBIN INSTITUTE OF TECHNOLOGY (SHENZHEN) (INSTITUTE OF SCIENCE AND TECHNOLOGY INNOVATION HARBIN INSTITUTE OF TECHNOLOGY SHENZHEN)
Filing Date
2024-12-26
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

Existing distributed training methods for models lack reliability when aggregating global models. This is mainly because the trust management mechanism is limited to the blockchain and does not consider local training behavior data of nodes, resulting in a high risk of malicious nodes and data tampering, which affects the training quality of the global model.

Method used

By using the local model training behavior data of each computing node to determine the trust value during the distributed training process, and aggregating the global model based on trust management, including node registration, trust value calculation, penalty mechanism and global model generation, the immutability and distributed storage of blockchain are used to ensure data integrity and security.

Benefits of technology

It improves the security and reliability of distributed training computing nodes, enhances the training quality of the global model, prevents malicious node attacks, and ensures the transparency and integrity of data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120012168B_ABST
    Figure CN120012168B_ABST
Patent Text Reader

Abstract

The application provides a model distributed training method based on blockchain node trust management, and the method comprises the following steps: a computing node trains an initial model according to a training data set, and determines local training behavior data; the computing node determines a training trust value according to the local training behavior data; the computing node trains the initial model according to the training data set and the training trust value; wherein, when the initial model training is completed, a master node obtains the trained model from a slave node, aggregates the trained model to generate a global model, and sends the global model to a user end. The model of the computing node with a high trust value is aggregated to generate a global model through the model training of the computing node based on the trust management according to the local training behavior data, so as to improve the security and reliability of the distributed training computing node, and further improve the training quality of the global model.
Need to check novelty before this filing date? Find Prior Art