|
Work item:
|
F.FFM
|
|
Subject/title:
|
Requirements and Framework for Federated Foundation Model
|
|
Status:
|
Under study
|
|
Approval process:
|
TAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2027-12 (Medium priority)
|
|
Liaison:
|
ITU-T SG17
|
|
Supporting members:
|
Institute of Computing Technology Chinese Academy of Sciences, China Telecommunications Corporation, China Academy of Information and Communications Technology, China Unicom
|
|
Summary:
|
The effectiveness of large model depends heavily on the model size and the scope of the training dataset. However, due to privacy protection, large amounts of private data and resources are difficult to be used in large model’s development. Federated foundation model offers a flexible and scalable framework that integrates federated learning into the entire lifecycle of large model in a privacy-preserving manner, enabling distributed collaborative training of large model and fostering more efficient, personalized, and privacy-conscious AI systems.
This Recommendation introduces the basic concepts of federated foundation model, identifies the requirements and framework, and specifies the corresponding training paradigms and tasks. Use cases and application scenarios are provided in the appendix to guide the application of federated foundation model, serving as a general reference for federated training of large model.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2025-09-08 16:09:39
|
|
Last update:
2026-05-11 15:56:19
|
