FRG Biweekly Meeting: Shahar Kovalsky and Fengyu Yang, UNC Math

2:00-2:10pm Lighting Talk by Fengyu Yang

2:10-3:00pm Informal talk by Prof. Shahar Kovalsky

Title: Two ongoing Projects (tentative)

Abstract: In this talk, we will discuss two ongoing projects. The first project, led by Connor Magoon, focuses on developing neural networks designed to compute well-behaved mappings of triangular meshes. In this work, we design a differentiable neural layer that addresses the challenge of ensuring that a piecewise linear mapping has bounded conformal distortion. Our approach builds upon prior work that used conic convex optimization to enforce geometric constraints (some of which were mentioned in our prior presentation; we promise to provide a brief refresher). The second project, led by Fengyu Yang, is concerned with the joint analysis of two modalities of brain imaging data: structural connectome data, extracted from diffusion tensor imaging, and temporal sequences of blood-oxygenation-level-dependent signals. Through spectral analysis of this data, we investigate brain interactions in both space and time. Our ultimate objective is to identify activity patterns that hold potential for early diagnosis of degenerative diseases.

Note: This talk will not be recorded.