|
Work item:
|
F.TCEF-FML
|
|
Subject/title:
|
Trusted contribution evaluation framework on federated machine learning services
|
|
Status:
|
[Carried to next study period]
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
-
|
|
Liaison:
|
ITU-T SG13, SG20
|
|
Supporting members:
|
China Unicom; China Information Communication Technologies Group; ZTE Corporation; ETRI
|
|
Summary:
|
Federated machine learning (FML) is an emerging distributed framework that enables collaborative machine learning (ML) and model construction across decentralized datasets on the basis of ensuring data security and private and legal compliance. In FML, where the computing for machine learning is where the data. FML allows participants to jointly training on the basis of not sharing data, which can technically break data islands and achieve collaborations.
FML involves multiple participants, and each participant's contribution to the training results usually is different. Contribution degree on FML service is used to measure the contribution of different participant to the final FML result. Participant with high contribution degree deserve higher award. An effective and reliable evaluation mechanism for contribution degree on FML service is essential for the motivation of current and potential FML participants and can promote the sustainable development of FML services.
This draft new Recommendation introduces an evaluation service for contribution degree on federated machine learning service, and provides its concept, characteristics, requirement, use cases, and specifies its reference framework and common capabilities.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2021-05-21 11:58:01
|
|
Last update:
2022-02-15 13:35:07
|
