| |
|
| |
|
Mark Sandler
Cornell University
Algorithms for Mixture Models: Analysis and Experiments
April 27, 2006 10:00am
Abstract:
Mixture models form one of the most fundamental classes of generative models for clustered data, and they have numerous applications in information retrieval, computer vision, and other fields. However, until very recently, there have been only a few algorithms known that can provably guarantee reconstruction of the model with small error, given observed data. In this talk we present algorithms for two different learning problems which enjoy this property. We also present results of experiments performed on collection of abstracts from ArXiV using one of our algorithms.
One interesting property shared by both our algorithms is that they are based on the use of the L1-norm, and we show that it is a better choice than Euclidean norm for these problems.
If you have questions, or would like to meet the speaker, please contact Ponda at 4-1994 or pondabarnes@tti-c.org. For information on future TTI-C talks or events, please go to the TTI-C Events page.