Linear and quadratic matching appear in countless computing disciplines, from algorithmic theory to applications in shape analysis, machine learning, and medical imaging. Motivated by special properties of these problems when they are associated with a geometric domain, in this talk I will discuss practical tools for efficient approximation of solutions to matching problems. In particular, I will discuss model problems from the field of optimal transport, showing how to leverage geometric structure and regularization to develop fast, easily-implemented algorithms that are readily incorporated into the geometry pipeline.
Structured Assignment: Practical Tools for Applied Linear and Quadratic Matching
Apr 12 2016 - 12:30pm
Graduate Student Seminar
Fine Hall 214