In person & onlineAI/ML in 5G

ITU AI/ML Challenge launch: Application inference from packet flows

  • * Register (or log in) to the Neural Network to add this session to your agenda or watch the replay

  • Date
    10 September 2026
    Timeframe
    11:00 - 12:00 CEST Geneva
    Duration
    60 minutes

      This webinar introduces the ITU AI/ML Challenge on application inference from encrypted packet flows. The challenge asks participants to classify mobile application flows using header-derived traffic statistics, without relying on payload, IP addresses, DNS names, TLS SNI, HTTP Host values, URLs, or domain strings. The webinar will explain the motivation, dataset structure, two-task setup, submission format, evaluation metrics, and important constraints.

      After the session, participants will understand the goal of the challenge, how to use the released training and validation data, how to prepare submissions for both the static full-flow task and the real-time first-5-packet task, and how submissions will be evaluated using macro F1 and accuracy.

      Are you sure you want to remove this speaker?