Reproduce “Geometric deep learning on graphs and manifolds using mixture model 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
Federico Monti, Davide Boscaini, Jonathan Masci, Emanuele Rodolà, Jan Svoboda, Michael M. Bronstein
MoNet provides a learnable local patch operation for graph and surface data, using domain-specific pseudo-coordinates to weight information from neighboring points. It can express several earlier graph and manifold convolutional methods as special cases. On MNIST images represented by 150-superpixel graphs, MoNet reaches 96.75% test accuracy versus 80.94% for ChebNet; on shared regular grids, results are approximately equal.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.