Reproduce “Subspace Robust Wasserstein Distances” 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
François-Pierre Paty, Marco Cuturi
The authors define subspace-robust Wasserstein (SRW) transport, which chooses a coupling to minimize the sum of the k largest eigenvalues of its displacement matrix, rather than total squared movement. In a dimension-20 experiment with rank-5 Gaussian supports and added noise, SRW showed smaller relative changes than ordinary Wasserstein at moderate-to-high noise; at low noise, computational imprecision affected results.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.