HodgeNet
Abstract
Constrained by the limitations of learning toolkits engineered for other applications, such as those in image processing, many mesh-based learning algorithms employ data flows that would be atypical from the perspective of conventional geometry processing. As an alternative, we present a technique for learning from meshes built from standard geometry processing modules and operations. We show that low-order eigenvalue/eigenvector computation from operators parameterized using discrete exterior calculus is amenable to efficient approximate backpropagation, yielding spectral per-element or per-mesh features with similar formulas to classical descriptors like the heat/wave kernel signatures. Our model uses few parameters, generalizes to high-resolution meshes, and exhibits performance and time complexity on par with past work.
Document Details
- Document Type
- Pub Defense Publication
- Publication Date
- Jul 19, 2021
- Source ID
- 10.1145/3450626.3459797
Entities
People
- Dmitriy Smirnov
- Justin Solomon
Organizations
- Adobe
- Air Force Office of Scientific Research
- Army Research Office
- Massachusetts Institute of Technology
- National Science Foundation
- Toyota Motor North America