U of M: MCIM, School of Mathematics

2005-2006 IMA/MCIM Industrial Problems Seminar
An IMA/MCIM Joint Seminar in Applied Mathematics Fridays at 1:25pm in 570 Vincent Hall
MCIM students will give talks on their internships and research on the Fridays when there are no outside industrial speakers scheduled. If you would like to sign up to give a talk, or make an appointment to speak with any of the industrial speakers listed below please call 612-625-3377 or e-mail Rhonda
2005 - 2006
September 30, 2005	Assaf Naor, Microsoft Graph Partitioning, Clustering with Qualitative Information, and Grothendieck-type Inequalities In this talk we will describe applications of Grothendieck's inequality to combinatorial optimization. We will show how Grothendieck's inequality can be used to provide efficient algorithms which approximate the cut-norm of a matrix, and construct Szemeredi partitions. Moreover, we will show how to derive a new class of Grothendieck type inequalities which can be used to give approximation algorithms for Correlation Clustering on a wide class of judgment graphs, and to approximate ground states of spin glasses. No prerequisites will be assumed, in particular, we will present a proof of Grothendieck's inequality.
October 7, 2005	Tin Kam Ho, Bell Labs, Lucent Technologies Geometrical complexity of classification problems Pattern recognition seeks to identify and model regularities in empirical data by algorithmic processes. Successful application of the established methods requires good understanding of their behavior and also how well they match the application context. Difficulties can arise from either the intrinsic complexity of a problem or a mismatch of methods to problems. We describe some measures that can characterize the intrinsic complexity of a classification problem and its relationship to classifier performance. The measures revealed that a collection of real-world problems can span an interesting continuum between those easily learnable to those with no learning possible. We discuss our results on identifying the domains of dominant competence of several popular classifiers in this measurement space.
October 14, 2005	Jeff Remillad, Ford
October 28, 2005	Valerio Pascucci, LLNL
November 4, 2005	David Arathorn, General Intelligence Corp.
November 18, 2005	TBA
December 2, 2005	TBA

Industrial Problems Seminar 2004-05
Industrial Problems Seminar 2003-04
Industrial Problems Seminar 2002-01