Learning Smooth Neural Functions via Lipschitz Regularization
Event Type
Technical Paper
Interest Areas
Research & Education
Presentation Types
Registration Categories
Full Conference Supporter
Full Conference
Virtual Conference Supporter
Virtual Conference
Exhibitor Additional Full Conference
Exhibitor Full Conference
DescriptionSmoothness is an essential property in geometry processing with neural fields. We introduce a Lipschitz regularization to encourage smoothness in functions parameterized by neural networks. Ours is computationally fast and can be implemented in four lines of code. We demonstrate the effectiveness on shape interpolation, extrapolation, and latent code optimization.