We are pleased to share that we have submitted an application to host the BINGO Challenge as a sponsored competition on the Kaggle platform and are currently awaiting the outcome of the review process!
If approved, this competition would invite data scientists, machine learning practitioners, neuroscientists, and curious innovators from around the world to address one of the most challenging problems in brain–computer interface research: EEG-based imagined speech decoding. Participants would be provided with a rich multi-session EEG dataset collected during the imagined production of the full NATO phonetic alphabet, with the aim of developing robust algorithms capable of accurately classifying imagined speech patterns — a key step toward future assistive communication technologies.
By proposing the BINGO Challenge as a Kaggle-sponsored competition, we aim to leverage a well-established environment for transparent benchmarking, reproducible evaluation, and collaborative learning. The planned competition setup includes baseline models, clearly defined training and test splits, and standardised evaluation metrics. The initiative is intended to be accessible to academic researchers, industry practitioners, and early-career machine learning enthusiasts alike.
Why this matters:
- Open-access benchmark: A comprehensive imagined speech EEG dataset made available to the community.
- Fair and transparent evaluation: Standard metrics and fixed data splits to enable meaningful comparison.
- Community engagement: Opportunities for discussion, collaboration, and learning through Kaggle’s forums and leaderboards.
- Potential societal impact: Advancing research toward assistive communication technologies for people with severe motor impairments.
We will share updates as soon as a decision is made. In the meantime, we look forward to engaging with the community around this exciting research direction.