一种自适应客户端参数更新的联邦学习方法、系统及介质
By employing adaptive classification and performance testing, the problem of client device mismatch in federated learning is resolved, thereby improving the accuracy and efficiency of model training.
CN117787442BActive Publication Date: 2026-07-17YIJIAN (SHANGHAI) INFORMATION TECH CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- YIJIAN (SHANGHAI) INFORMATION TECH CO LTD
- Filing Date
- 2023-12-27
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing technologies lack adaptive selection and timely detection mechanisms, leading to performance mismatches between client devices and federated learning, which affects the accuracy of model training.
Method used
By acquiring information on the client's basic performance and model training requirements, adaptive classification is performed to detect and replace client devices that do not meet the conditions for continued training, thus achieving adaptive update selection.
Benefits of technology
It enables precise training based on model training requirements and device performance, improving the accuracy and efficiency of model training.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN117787442B_ABST
Abstract
本申请提供了一种自适应客户端参数更新的联邦学习方法、系统及介质。该方法包括：该方法包括：根据客户端本地数据的类别、服务器模型训练需求、客户端基本性能处理获得参与度指数，并上传至服务器，并根据参与度指数等级将所有客户端划分为训练客户端和备选客户端，训练客户端对初始模型进行训练，获得模型更新参数并加密后上传至服务器，对训练客户端的训练过程和服务器后台分别进行监测并进行性能评估，选取符合性能评估要求的客户端作为候选训练客户端，服务器将所有训练客户端的加密模型更新参数解密后进行聚合处理，生成全局模型，并下发至候选训练客户端进行下一轮训练。本申请可以实现对参与训练的客户端进行自适应更新选择的目的。
Need to check novelty before this filing date? Find Prior Art