Reproduce “Geo2Vec: Shape- and Distance-Aware Neural Representation of Geospatial Entities” 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
Chen Chu, Cyrus Shahabi
Geo2Vec learns embeddings for geographic points, lines, and building footprints by modeling signed distance fields in map coordinates. It samples near vertices, along edges, and across the space, then trains a neural network to predict distances to boundaries. In Singapore polyline relation classification, it achieves 96.4% accuracy versus Poly2Vec’s 81.2%, using 256-dimensional embeddings; this result is limited to that task.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.