Return to Colloquia & Seminar listing
Geometry and Statistical Learning
Applied Math| Speaker: | Misha Belkin, Ohio State University |
| Location: | 1147 MSB |
| Start time: | Mon, Oct 23 2006, 4:10PM |
I will discuss why geometry of high-dimensional data may be useful for various problems of statistical learning. In particular, I will talk about the role of the Laplace operator on a manifold, explain how it may be estimated from point-cloud data, when the underlying manifold is not known, and present some resulting algorithms. I will also discuss connections to the heat equation and show some convergence results.