Reproduce “GNN-VIEW: Visual Inspection and Explanation of Graph-based Earth-System Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Alessandro Lucatelli, Ferdinand Unterhuber, Katharina Krenzer, Mennatallah El-Assady, Benedikt Soja, Christina Humer, Joel Oskarsson
GNN-VIEW lets developers inspect graphs before training and trace inputs behind selected Earth-system forecasts after training. Its Explore and Inspect modes expose graph connections, while Explain maps integrated-gradients attributions across variables, times, and locations. In Hurricane Laura precipitation, attribution shares correlate +0.57 with field-deletion prediction changes, versus +0.17 for input magnitude; this is evidence, not proof of physical causes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.