Reproduce “Laplace Learning in Wasserstein Space” 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

Mary Chriselda Antony Oliver, Michael Roberts, Carola-Bibiane Schönlieb, Matthew Thorpe

Laplace Learning extends semi-supervised labeling to distribution-valued data, representing clouds as probability measures and propagating known labels across a transport-weighted graph. The authors prove almost-sure Γ-convergence of graph energies under stated assumptions. On ModelNet10, they report an advantage over supervised PointNet, but caution the paradigms are not strictly comparable; experiments use a linear Wasserstein approximation rather than 2-Wasserstein distance.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.