Reproduce “Bayesian Sheaf Neural Networks” 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

Patrick Gillespie, Layal Bou Hamdan, Ioannis Schizas, David L. Boothe, Vasileios Maroulas

Bayesian Sheaf Neural Networks (BSNNs) model message-passing structures as distributions over sheaves rather than single choices. They sample sheaves to form Laplacian-based updates and train an approximate distribution with a variational objective containing a KL penalty. Across datasets, BSNNs are competitive but do not always lead; in limited-data tests, Diag-BSNN outperforms Diag-NSD on Texas, 73.48% versus 71.22% accuracy.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.