Phase Transitions, Distance Functions, and Implicit Neural Representations

                            Yaron Lipman 1 2

             Abstract
   Representing surfaces as zero level sets of neural
   networks recently emerged as a powerful mod-
   eling paradigm, named Implicit Neural Repre-
   sentations (INRs), serving numerous downstream
   applications in geometric deep learning and 3D vi-
   sion. Training INRs previously required choosing
   between occupancy and distance function repre-
...