ML Challenge: Build-a-thon 3.0 for AI Native Networks and Services

Introduction

Description

1. Background:

Artificial Intelligence (AI) is poised to become the predominant technology in the future, influencing every facet of our society. On one hand AI / Machine Learning (ML) is set to enable new applications and services redefining the end-user experience, and on the other hand AI/ML enabled optimization techniques would simplify operations of communication networks. International Telecommunication Union (ITU) has taken a leading role in this pursuit, actively studying applications of AI/ML in future networks, including 5G and 6G. This commitment is exemplified by ITU's initiatives such as the AI/ML 5G Challenges, which provide datasets and problem statements, ITU Journal for Future and Evolving Technologies, AIforGood Discovery series of webinars, and ML5G mentoring programs. ITU published ITU-T Y.3172 which specifies an architectural framework for machine learning in future networks including IMT-2020. The ML pipeline and ML management and orchestration functionalities (MLFO) defined in Y.3172 enabled a common vocabulary for integration of AI/ML components into 5G and future networks Building upon ITU-T Y.3172 ITU launched ML5G Challenges in 2020. The AI/ML5G Challenge is a platform aimed at addressing the challenges of implementing AI/ML in communication networks, including 5G and upcoming 6G technologies. Participants from over 100 countries collaborated to solve real-world problems using AI/ML techniques, aligned with ITU standards. The Challenge provides carefully selected problem statements, a combination of open data from real-world and simulations, supported by technical webinars from experts, mentoring, and practical hands-on sessions. The participating teams work on exploring the data, creating, training, and deploying ML models tailored for communication networks. Participation in the Challenge allows individuals to showcase their skills, test concepts using real data and problems, compete for global recognition, prizes, and certificates. Moreover, participants can also align their solutions with ITU standards, thus potentially entering the realm of international telecommunications standards. One of the important applications of AI/ML in networks is to enable AI Native Networks. ITU established FG AINN (Focus Group on AI Native Networks) which studies use cases, architecture and Proof of concepts related to AI Native Networks. India holds many leadership positions in the group, including the Vice Chair position. Build-a-thon was launched by ITU as a coding competition, in collaboration with ITU ML5G Challenge and FG AINN to create reference implementation of use cases and architecture components. This initiative was supported by several online workshops and mentoring by ITU FG experts. Teams around the world competed and won prizes as part of the Build-a-thon initiative. Continuing the momentum created by these activities, ITU is conducting Build-a-thon 3.0 during IMC 2025 in New Delhi. Once again, the theme is to apply ITU standards and create applications which integrate AI/ML, and are locally and globally relevant. The Build-a-thon will be announced widely in ITU channels and online preparatory and mentoring sessions would be held till the actual event.

There would be two phases:

• Phase-1 online preparatory sessions would be open till Oct 2025.
• Phase-2 is an in-person coding event on 9th October 2025. This contest would culminate in a prize distribution ceremony during IMC.

2. Abstract for problem statement:

AI Native Bharat 5G – Build Your Own AI/ML Model for 5G

• Description: In this problem statement, the participants will come together, undergo specific trainings and mentoring for ITU Recommendations on AI/ML. They will pick up specific usecases of their choice from a list of 5G/6G usecases, and utilize the resources populated in the Sandbox to design and build AI/ML pipelines and agents corresponding to the use case. The resources provided to the participants may include open data, compute servers, and mentoring by ITU experts.

3. Phases

Phase-1: Preparatory phase: setup Sandbox.

This includes populating data, storage and compute. This could be in cloud. Reference ITU-T Y.3172, Y.3181, FG AINN documents. The following would be conducted during this phase: o Webinars, roundtables, mentoring. o call for 6G usecases, could be application or core usecase o Design ML pipelines, o Identify open data. o Model training

Phase-2 (in the room)

Demo in Sandbox

4. Incentives:

First prize: INR 20000/-
Second prize: INR 15000/-
Third prize: INR 10000/-
Prize for best girls team: INR 10000/-
Mentors’ nomination for best engaging team: INR 10000/-

Contact

AI-5G-Challenge, ITU AI5GChallenge@itu.int