Page 24 - Crowdsourcing AI and Machine Learning solutions for SDGs - ITU AI/ML Challenges 2024 Report

Crowdsourcing AI and Machine Learning solutions for SDGs

6.2 Build-a-thon

The build-a-thon is a problem statement in ITU AI/ML in the 5G Challenge hosted by ITU-T
Focus Group Autonomous Networks (FG AN). It aims to demonstrate and validate important
use cases for autonomous networks, creating proof of concept implementations and tools in
the process. As an open platform, FG AN is well-poised to enable access to experts, students,
and industry, collaborate with other external events such as plugfests and hackathons, and
set the stage for collaboration in open-source projects, other proof of concept (PoC) and
standardization work.

In essence, a build-a-thon is a common, open platform for like-minded people to come together
(remotely) and build something to prove a point. In the case of FG AN, it is defined as below:

1 Build-a-thon is a PoC development activity, to build upon a key concept in FG AN,
especially intended to prove the concept practically with code, test setup, and demo
setup.
2 Build-a-thon is not intended to create a product, nor would the code created as part of
Build-a-thon be considered as product quality software.
3 Build-a-thon would create well-documented artifacts and open-source code.

Winning solutions were showcased during build-a-thon workshops, and some contributions
were integrated into the development of FG-AN deliverables. These deliverables formed the
basis of SG13 recommendations and technical reports on autonomous networks, such as
Y.3061.

The build-a-thon problem statements: https:// challenge .aiforgood .itu .int/ match/ matchitem/ 68

The Proof-of-concept report by FG AN summarizes the achievements of the build-a-thon
activities: https:// www .itu .int/ en/ ITU -T/ focusgroups/ an/ Documents/ PoC _activities .pdf

6.3 Graph Neural Networks (GNN)

The Graph Neural Networks challenge, organized from 2020 to 2023, focused on leveraging
GNN for various applications within communication networks and beyond. Participants
developed innovative solutions using GNN techniques, contributing to the advancement of
this cutting-edge field. The challenge provided a platform for exploring the potential of GNN
in improving network performance and other complex tasks.

Description and details of different editions of the Graph Neural Networks challenge can be
found here: https:// bnn .upc .edu/ challenge/

Workshops: https:// bnn .upc .edu/ workshops/ gnnet2024/
Paper: The Graph Neural Networking Challenge: AWorldwide Competition for Education in AI/
ML for Networks: https:// dl .acm .org/ doi/ 10 .1145/ 3477482 .3477485

16