2025 NSF HDR Scientific Modeling Out-of-Distribution Challenge

Achieved 2nd Place at the 2025 NSF HDR (Scientific-Mood) Hackathon in Taiwan in the iHARP Track (Predicting Coastal Flooding Events).

Global Leaderboard Highlights (Team NCKU_UCLab):

• Global Tied 3rd Place in the iHARP Track (Predicting Coastal Flooding Events).
• Global Tied 4th Place in the Overall Ranking across all challenge tracks.

Awarded Project: “Coastal Flooding Events Prediction” In this challenge, our team developed robust machine learning models (integrating approaches like XGBoost and spatio-temporal data analysis) designed for out-of-distribution scientific forecasting. Our methodology demonstrated strong predictive performance for complex coastal flooding events, securing top rankings on both regional and global leaderboards.