US Patent Application 18331929. METHOD FOR SEMI-ASYNCHRONOUS FEDERATED LEARNING AND COMMUNICATION APPARATUS simplified abstract

From WikiPatents
Jump to navigation Jump to search

METHOD FOR SEMI-ASYNCHRONOUS FEDERATED LEARNING AND COMMUNICATION APPARATUS

Organization Name

HUAWEI TECHNOLOGIES CO., LTD.


Inventor(s)

Zhaoyang Zhang of Hangzhou (CN)


Zhongyu Wang of Hangzhou (CN)


Tianhang Yu of Hangzhou (CN)


Jian Wang of Hangzhou (CN)


METHOD FOR SEMI-ASYNCHRONOUS FEDERATED LEARNING AND COMMUNICATION APPARATUS - A simplified explanation of the abstract

  • This abstract for appeared for US patent application number 18331929 Titled 'METHOD FOR SEMI-ASYNCHRONOUS FEDERATED LEARNING AND COMMUNICATION APPARATUS'

Simplified Explanation

This application introduces a method for federated learning, which addresses the challenges of low training efficiency, unstable convergence, and poor generalization capability in both synchronous and asynchronous systems. It achieves this by setting a threshold for triggering the fusion of local models sent by terminal devices to generate a global model. The fusion weight of the local model is determined by considering factors such as data features, lag degree, and utilization degree of the sample set. By doing so, the application avoids the need for synchronizing model uploading versions in a synchronous system and overcomes the limitations of the "update upon reception" principle in an asynchronous system.


Original Abstract Submitted

This application provides a method for federated learning. A communication apparatus triggers, by setting a threshold (a time threshold and/or a count threshold), fusion of a local model sent by a terminal device, to generate a global model, and when a fusion weight of the local model is designed, a data feature included in the local model of the terminal device, a lag degree, and a utilization degree of a data feature of a sample set of the corresponding terminal device are comprehensively considered, so that a problem of low training efficiency caused by a synchronization requirement for model uploading versions in a synchronous system can be avoided, and a problem of unstable convergence and a poor generalization capability caused by an “update upon reception” principle of an asynchronous system can be avoided.