A Model-Agnostic Physics-Guided Adapter for Few-Shot Transfer of Coastal Flood Prediction Models to Unseen Regions

License: See LICENSE file · Updated 2026-09-29
My rating

The paper targets high-resolution coastal flood prediction using deep learning surrogates for hydrodynamic simulation. It introduces the Physics Adapter, a compact, architecture-agnostic adaptation interface designed for few-shot transfer to unseen coastal regions. The method addresses the cost of generating target-region data for fine-tuning by reducing dependence on extensive numerical simulations.

Tags: