Work item:
|
Y.CSDL
|
Subject/title:
|
Requirements and framework for crowdsourced system based on distributed learning
|
Status:
|
Under study
|
Approval process:
|
TAP
|
Type of work item:
|
Recommendation
|
Version:
|
New
|
Equivalent number:
|
-
|
Timing:
|
2025-Q2 (Medium priority)
|
Liaison:
|
ITU-T SG16
|
Supporting members:
|
China Unicom, Huazhong University of Science & Technology, China Mobile, ZTE
|
Summary:
|
Building up distributed learning mode for crowdsourced system and enhance the crowdsourced service data security is important. The distributed learning mode can effectively solve the data security problem in the crowdsourcing process. The crowdsourced raw data will be computed on the distributed worker node and only encrypted intermediate computation result will be transferred to the crowdsourced platform. Besides, compared with the traditional crowdsourced systems utilizing the centralized computation mode, which was defined in ITU-T Y.4217, the realization of crowdsourced system based on distributed learning can also provide intelligent cooperation capabilities between crowdsourced resources and reduce bandwidth consumption and transmission time consumption.
This recommendation specifies the requirements and framework for crowdsourced system based on distributed learning in order to solve the distributed model parallelism and data parallelism issues for complex crowdsourced task, crowdsourced nodes communication strategy under distributed learning, coordination mode of heterogeneous models and data.
|
Comment:
|
-
|
Reference(s):
|
|
|
Historic references:
|
Contact(s):
|
|
ITU-T A.5 justification(s): |
|
|
|
First registration in the WP:
2023-02-16 11:44:24
|
Last update:
2023-09-27 18:08:11
|