Mathematics Colloquia and Seminars

Return to Colloquia & Seminar listing

Topological Methods in Data Analysis


Speaker: Gunnar Carlsson, Stanford University
Location: 1147 MSB
Start time: Tue, May 22 2007, 4:10PM

Algebraic topology is an effective formalism for studying geometric objects. It provides signatures which effectively encode many of our intuitive concepts about the qualitative nature of spaces, such as breaking up into components, possessing loops, possessing higher dimensional holes, etc. Recently, a methodology called persistent homology has been developed which permits the transport of these methods from the world of pure geometry to the probabilistic world of point cloud data, i.e. large but finite point sets sampled from a geometric object. I will discuss this theory, and show how it plays out in some examples.

Refreshments will start at 3:45 in MSB 1147. Dinner will follow the Colloquium.