Reproduce “Manifold Topology Divergence: a Framework for Comparing Data Manifolds” 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
Serguei Barannikov, Ilya Trofimov, Grigorii Sotnikov, Ekaterina Trimbach, Alexander Korotin, Alexander Filippov, Evgeny Burnaev
MTop-Div compares generated and reference samples across scales using persistent homology, rather than assessing each sample cloud separately. In controlled CIFAR10 tests of several disturbances, its average Kendall rank correlation with disturbance severity was 0.89, versus 0.36 for Geometry Score. This supports sensitivity to those tested disturbances, not human-perceived quality or universal superiority.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.