Reproduce “Uncovering Challenges of Solving the Continuous Gromov-Wasserstein Problem” 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
Xavier Aramayo Carrasco, Maksim Nekrashevich, Petr Mokrov, Evgeny Burnaev, Alexander Korotin
The authors test matching word embeddings across 100- and 50-dimensional spaces as training pairings are removed. Existing baselines’ matching performance declines. NeuralGW learns a pairing-free, sample-based inner-product Gromov–Wasserstein map and performs best in tested large-scale unpaired settings, but varies substantially with initialization and requires plenty of data, so this does not establish a reliable general-purpose solver.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.