Systematically differentiating parametric discontinuities
Abstract
Emerging research in computer graphics, inverse problems, and machine learning requires us to differentiate and optimize parametric discontinuities. These discontinuities appear in object boundaries, occlusion, contact, and sudden change over time. In many domains, such as rendering and physics simulation, we differentiate the parameters of models that are expressed as integrals over discontinuous functions. Ignoring the discontinuities during differentiation often has a significant impact on the optimization process. Previous approaches either apply specialized hand-derived solutions, smooth out the discontinuities, or rely on incorrect automatic differentiation.
Document Details
- Document Type
- Pub Defense Publication
- Publication Date
- Jul 19, 2021
- Source ID
- 10.1145/3450626.3459775
Entities
People
- Gilbert Bernstein
- Jesse Michel
- Jonathan Ragan-Kelley
- Kevin Mu
- Sai Praveen Bangaru
- Tzu-mao Li
Organizations
- Defense Advanced Research Projects Agency
- Massachusetts Institute of Technology
- University of California, Berkeley