Neurotech · Draft-source example
Connectomics Benchmark + Prize Program
Make connectome reconstruction a measurable, competitive AI problem.
The bottleneck
Proofreading and reconstruction lack an operationally maintained benchmark, blinded evaluation set, and cost curve.
The proposed work
Blinded datasets, reproducible evaluation, a maintained leaderboard, and a recurring Connectome Challenge.
PL R&D’s proposed role
PL’s role is to be confirmed. The source proposes a field-level program but does not establish PL’s specific delivery responsibility.
What we would examine
Measure accuracy, compute cost, and human labor; the draft targets serious outside participation and order-of-magnitude improvement.
How to get involved
Production pages should offer a reviewed, program-specific route to contribute expertise, partner, or discuss funding. No intake or outreach is connected here.
- Supporting types:
- Resource · Build
- Public timing:
- 18 months
- Classification:
- Suggested classification · needs FA review
This is a proposed public edition from internal planning drafts, not a verified active commitment. Funding figures and private source links are withheld until a designated publisher reviews them.