HIGHLIGHTS
- who: Tobias Alt from the Saarland University, Saarbru00fccken, Germany have published the paper: Designing rotationally invariant neural networks from PDEs and variational methods, in the Journal: (JOURNAL)
- what: The aim of the paper is to investigate how diffusion and variational models achieve and transfer these ideas to networks. As a core novelty the authors propose activation functions which couple network channels by combining information from several oriented filters. In the following, the authors explore other ways to design the energy functional based on rotationally invariant quantities and investigate how the resulting diffusion model changes. The . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.