Data-efficient flood depth prediction through domain-aware coreset selection and tabular foundation models
Researchers have developed a novel method for predicting flood depths more efficiently and accurately. This approach utilizes a domain-aware coreset construction pipeline that conditions a tabular foundation model during inference. By strategically sampling data based on storm characteristics and watershed impact, the model achieves high accuracy with significantly less training data, demonstrating strong transferability to new, unseen watersheds without retraining. AI
IMPACT This research offers a more data-efficient and transferable approach to flood prediction using foundation models, potentially improving disaster response and urban planning.