When: Monday, April 10, 4:00 p.m.- 5:30 p.m.

Where: 3305 Newell-Simon Hall

Aleix M. Martinez

VASC Seminar

Abstract:
Many problems in science and engineering can be formulated as a pattern recognition (or machine learning) problem. When using pattern recognition approaches, one prefers to employ linear methods-­mainly due to their simplicity and tractability, and because linear methods generally provide a closed form solution which can be made to work with small sample size datasets. Unfortunately, linear methods have many limitations, of which many are still unknown. Understanding these limitations is key to advance the current state of the art. In this talk, we will define when linear methods work, do not work and how this knowledge can be used to define algorithms that are guaranteed to work in a large number applications and under a large variety of assumptions. For simplicity, I will concentrate on the problems of feature extraction and classification. Several experimental results in vision, linguistics, bioinformatics, and data analysis will be shown.

<< Back