Kathryn Heal

Kathryn Heal is a PhD Candidate in applied mathematics at Harvard University, studying algebraic vision with Prof. Todd Zickler. Her work has focused on algebraic objects inspired by depth reconstruction problems within computer vision. Her current research considers the general solution set of an ill-posed problem called Shape from Shading as an algebraic variety. The algebro-geometric structure of this solution set yields nice properties that can be exploited for both a parsimonious computational representation for this object, and an efficient algorithm to solve the depth reconstruction problem. In addition to this area, she is also interested in geometric explanations for the overwhelming successes of neural networks as functional approximators. Her planned graduation date is Spring of 2020.