DINOZAUR
A diffusion multiplier for Fourier neural operators, with uncertainty
Paper title Light-Weight Diffusion Multiplier and Uncertainty Quantification for Fourier Neural Operators
Paper arXiv 2508.00643PDF
Replaces the dense spectral weight tensor in each Fourier layer with a heat-kernel diffusion multiplier that has one learnable diffusion time per channel, plus spatial-gradient features. Putting priors on those diffusion times turns the model into a Bayesian neural operator that samples spatially correlated fields.
- Published
- NeurIPS
Share on LinkedIn, X or Bluesky, or or .
Know related work that belongs here? Suggest a link
Something wrong on this page? Tell the editor