Reproduce “Möbius Convolutions for Spherical CNNs” 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
Thomas W. Mitchel, Noam Aigerman, Vladimir G. Kim, Michael Kazhdan
The Möbius-convolution network adapts spherical filters to conformal distortions by aligning each filter with the changing local frame and correcting for changes in spherical area. Its implementation approximates the transformed filter using truncation and quadrature. In the conformally augmented SHREC ’11 setup, it reaches 86.5% accuracy versus DiffusionNet’s 64.9%, but this result does not establish performance on real-world deformations.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.