Computer Science Research Institute

Title: Topological Data Analysis: Persistence & Tidy Sets

Speaker: Prof. Afra Zomorodian, Department of Computer Science, Dartmouth College

Date/Time: Tuesday, April 13, 2010, 3:00 - 4:00 pm

Location: CSRI Building/Room 279 (Sandia NM)

Brief Abstract: A real-world dataset may be modeled as a finite set of noisy points, sampled from some underlying space, but embedded in some high-dimensional metric space. Topological data analysis focuses on recovering the connectivity of the original space.

Topological analysis often has two steps:
1. approximating the underlying space with some combinatorial representation, and
2. computing topological invariants.

In this talk, I begin by motivating such analysis with an application to a problem in computer vision: discovering the local structure of natural images. I then describe some of my work on each of the two steps: persistence (step 2), and tidy sets (step 1).

CSRI POC: Scott Mitchell, (505) 845-7594