Reproduce “Gaussian 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
Peter Kuhn, Victoria Heusinger-Heß
The authors introduce Gaussian neural networks (GaNNs), which learn Gaussian priors over hidden-layer activities alongside a supervised task. An auxiliary loss scores how well activities fit those priors, adding an explicit regularization signal on representations. Across ten image and tabular datasets, sparse GaNNs outperform plain neural networks on nine and tie on one, though results are not uniformly better.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.