|
Work item:
|
F.VIS-FML
|
|
Subject/title:
|
Requirements and framework of visual inspection system based on federated machine learning in smart grid
|
|
Status:
|
Under study [Issued from previous study period]
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2026 (Medium priority)
|
|
Liaison:
|
ITU-T SG17; IEEE Power & Energy Society; ISO/IEC JTC 1/SC 42
|
|
Supporting members:
|
STATE GRID of China, MIIT, China Telecom, Nokia, Zhejiang Lab
|
|
Summary:
|
Visual object detection techniques are widely used in visual inspection systems in smart grid, visual object detection is a computer vision technique based on artificial intelligence (AI).
However, due to privacy concerns and the high cost of transmitting visual data (image and video), building object detection AI models based on centrally stored large training datasets is challenging. Data scarcity is another major challenge for high-quality AI model; training AI models needs a large amount of data to achieve exceptional performance. Unfortunately, many applications have small or inadequate datasets, especially in fault detection in smart grids.
Federated machine learning (FML) is a promising approach to resolve the above challenges, which enables multiple participants to collaboratively build and use machine learning models without disclosing the raw and private data owned by the participants while achieving good performance.
This draft new Recommendation defines the framework, requirements and use cases of the visual inspection system based on federated machine learning in smart grid.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2024-06-24 10:09:01
|
|
Last update:
2025-03-10 11:26:59
|
