NSF HDR Scientific Modeling Out-of-Distribution Challenge
Overview：
Co-hosted by three institutes under the U.S. National Science Foundation (NSF) Harnessing the Data Revolution (HDR) program: A3D3, Imageomics, and iHARP.
The competition focuses on the core challenge of Out-of-Distribution (OOD) prediction in scientific data.
This year’s challenge features three scientific tracks:
- Neural Forecasting: Predict future neuronal activity using electronic neuron signals recorded from monkeys
-
Climate Prediction using Ecological Data: Use ground beetle images to forecast drought trends and explore the impact of climate change on ecosystems.
- Coastal Flooding Prediction over Time: Predict future flood risks based on long-term sea-level data.
Continuing the spirit of the NSF Harnessing the Data Revolution (HDR) initiative — Open Science × Cross-Disciplinary AI — we invite researchers and AI experts worldwide to break through the data limitations of traditional models and push the boundaries of predictive performance in out-of-distribution (OOD) scientific scenarios.
Participation Guidelines
Experts and enthusiasts from all fields of AI are warmly welcome to join! Participants may compete individually or as teams in the hackathon challenges.
How to participate:
- Register on the Codabench platform and download the open-source datasets.
- Train and validate your models using the provided data.
- Upload your completed AI model to Codabench for evaluation.
Special for Taiwan Participants This Year!
We will have a Taiwan Regional Leaderboard and awards!
Global Prizes
-
💰 Cash Prize: USD $4,000
-
☁️ AWS & NVIDIA Cloud Computing Credits
-
✈️ Global Champion: Travel subsidy to attend the 2026 FARR Workshop in Washington, D.C., United States, for the global winner award ceremony and presentation
Taiwan Hackathon Session Details:
-
Date: Friday, December 19, 2025
-
Time: 12:00 noon
-
Format: Zoom intro session, followed by a local hackathon event with TAs, to check out the local event location in the event tab
-
Registration Link: Fill out the form https://forms.gle/2BjfAiLCDF9RWjKn7
-
The session will comprehensively cover the three challenge topics, data downloads, and submission guidelines.
-
Join and get help to submit your 1st fantastic ML models.
How to Register:
- Global official website: https://www.nsfhdr.org/mlchallenge-y2
NSF HDR Hackathon Key Timeline:
-
December 19, 2025: Online Hackathon Info Session + On-Site Hackathons at Various Taiwan Universities
-
January 7, 2026: Taiwan Regional leaderboard close date
-
January 14, 2026: NSF HDR Hackathon Taiwan Regional Award Ceremony
-
January 28, 2026: Global Overall leaderboard expected close date
-
April 8, 2026: Global FARR Conference Awards (International ceremony in the US
Co-Hosting Institutes：
 |
 |
 |
 |
 |
 |
 |
International Organizing Institute:
Sponsor：
|
|
 |
 |
 |
 |
 |
 |
International Co-Hosting Institutes：
