Call for abstracts is open — submit before 20 November 2026
byNumidia Institute of Technology

Big challenges · Day 2 · 17 December 2026

Team competitions on securing machine learning, reading the heart from 12-lead ECGs, recognising weeds in real fields and decoding the speech people hear from their EEG. Attack, defend, detect, classify or decode — then defend your work in front of the jury.
Big challenges
4
Ranked tracks
8
Open datasets
5
Challenge day & finals
17 Dec
How it works

  1. 01

    Form your team

    Students and professionals, alone or together — each challenge sets its own team size.

  2. 02

    Register it

    Create the team in your participant portal and share its code: teammates join in one click.

  3. 03

    Get the data

    Public datasets, starter notebooks, baselines and the organisers' own scoring scripts.

  4. 04

    Build & submit

    Upload predictions, model packages, code and reports from the team workspace. Your latest upload counts.

  5. 05

    Take the stage

    Finalists present to the jury on Day 2, and the winners are announced at the closing ceremony.

Key dates

The Auditory EEG call opens on 9 November with its preprocessing and scoring code, and its team registration closes on 27 November. Datasets and technical documentation for ECGDetect-AI and WeedDetect-AI are released on 13 November 2026. Final dates are confirmed by the organising committee.

  1. AI-Security Challenge — submissions close

    10 December 2026

  2. ECGDetect-AI 2026 — submissions close

    10 December 2026

  3. WeedDetect-AI 2026 — submissions close

    10 December 2026

  4. Auditory EEG Challenge — submissions close

    10 December 2026

  5. Final presentations, demonstrations and awards

    17 December 2026

Fair by design

Hidden or audited evaluation

AI-security tracks are scored on attacks and data teams never see, auditory EEG models are run on private recordings from unseen people, and ECG and weed models are checked through an audited, reproducible benchmark protocol.

Reproducible by default

Teams share their code and weights, and the organisers may re-run inference to verify any score.

Open, licensed data

PTB-XL and DeepWeeds are public under CC BY 4.0, and the auditory EEG datasets under non-commercial terms; the AI-security data and drone telemetry are generated by the organisers.

AI tools welcome, declared

Assistants are allowed. The challenges reward real understanding, and external data and pretrained models must be declared.

Ready?

Your team. Your track. Your stage.