Direct answers on eligibility, submission caps, audit, authorship, contingency, and Track 5 scope. Use this page before you submit.
Each answer is sourced from the competition proposal. If the FAQ contradicts something you read elsewhere on the site, treat this page as canonical and tell us. Email or Discord, same response time either way.
Multi-track entries are encouraged. Track 5 reuses the Tracks 1–4 evaluations through organizer-fitted heads, so a single foundation encoder gets you five leaderboard appearances.
Warm-up is generous. The sealed final tightens the cap so the prize roster reflects the model, not the search.
Yes. Tracks 1–4 accept independent submissions per team. Track 5 requires a single shared encoder; that encoder is automatically scored on Tracks 1–4 by organizer-fitted linear heads and appears on those leaderboards with a foundation-model label.
The warm-up phase (Jul 1 – Jul 31, 2026) is uncapped; iterate freely. The sealed final phase (Aug 1 – Sep 1, 2026 AoE) caps you at 5 submissions per team per day, so the prize roster reflects deliberate iteration rather than lottery search. The final NeurIPS ranking takes the best of your last five sealed submissions.
The competition has explicit fallbacks for low participation and delay. Neither outcome silently invalidates work already submitted; the rules below are how the organisers respond when reality disagrees with the plan.
Schedule shifts and analysis-group merges are announced on the website and Discord before they take effect.
Low-participation tracks fold into the closest active analysis group for the post-competition report, but prizes are still awarded on that track's own leaderboard. Track 5 is designed for this case: fewer Track-5 entries do not shrink the EEG-only field, because Tracks 1–4 are scored independently of Track 5 participation.
Technical, data, or infrastructure delays shift all downstream deadlines by the same amount on the website and Discord. The reproducibility audit gate is the only milestone that can extend independently if a top submission needs more re-run time.
Open by default. Organizer-affiliated entries appear on the leaderboard as reference but cannot take prize positions.
Industry, academia, students, independents: all eligible. Cross-institution teams are encouraged.
Anyone: industry, academia, students, independents. Cross-institution teams encouraged. Organizers and their direct lab members may submit, but their entries appear as Organizers · reference and are ineligible for cash prizes.
The PMLR competition report is the long-form artifact of the challenge. Reproducible top teams are co-authors. Pre-training data has a different rule per track: Tracks 1–4 prefer in-track training; Track 5 explicitly invites broad public-data use.
Pre-training corpora and compute estimates go in the method description. The audit reads them; the report names you.
Top-ranked teams that pass the reproducibility audit and submit a method description, training and inference code, and pre-training disclosures are invited as named authors on the PMLR competition report. Consortium authorship is offered to teams with complete, reproducible artifacts.
For Tracks 1–4, training data is suggested to stay within the track, but pre-training on any publicly available dataset is allowed; reproducibility scripts from top submissions are checked at audit. For Track 5, any public dataset is explicitly encouraged. The sealed test split is never allowed; closed clinical datasets are not allowed. Declare every external corpus in your method description.
Track 5 is the cross-modality stress test. A single set of encoder weights, evaluated through organizer-fitted heads on all four tasks. The two questions below mark the boundary between permitted adaptation and forbidden per-track tuning.
Manual code review at audit verifies that the same weights serve every task. Hidden-data fine-tuning is forbidden across the board.
Foundation Transfer tests whether a single encoder reuses across modalities, not just across EEG tasks. EMG is the cross-modality stress test. Without it, Track 5 would only measure multi-task EEG, which the 2025 challenge already covered.
Per-task linear heads and organizer-fitted probes are allowed. Per-track fine-tuning of the encoder weights is not. The encoder must remain a single set of weights, verified by manual code review at audit. Re-training or fine-tuning on hidden evaluation data is forbidden across all tracks.
New data is released where consent and licensing allow. Hidden evaluation labels stay confidential indefinitely. The audit re-runs the submitted training pipeline and compares scores against a tolerance band.
Arnaud Delorme (EEGLAB) chairs the reproducibility audit. Tolerance band is ±2σ around the submitted score.
After the challenge, new data is made publicly available in full or in part where consent, licensing, and provider policy allow. The hidden evaluation labels remain confidential indefinitely so the benchmark stays usable for future research.
The audit re-runs your training pipeline from the committed config and re-scores against the sealed split. Scores within ±2σ of your submitted number stay on the prize roster; scores outside the tolerance drop off the prize roster but remain on the public board for context. The audit is led by Arnaud Delorme (EEGLAB).
Email or Discord. Same response time either way.