[CCoE Notice] Industrial Engineering Seminar

[Tekin, Eylem]

INDUSTRIAL ENGINEERING SEMINAR

Date: October 11, 2013
Time: 10am-11am
Place: L2D2

The Interplay between Data Clustering and Optimization
Dr. Jiming Peng
Department of Industrial Engineering
University of Houston

Abstract
Clustering is one of the major tasks in data mining and knowledge discovery that has been widely used in a broad range of applications from customer segmentation to biological discovery.  Optimization is a field that deals with the problem of minimizing an objective function subject to various constraints over the variables. In this talk, we focus on the interaction between data clustering and modern optimization, and illustrate how the cross-fertilization between these two fields can help to advance the study in each of these two fields, respectively. We start by showing how several well-known clustering models such as k-means clustering and normalized cut can be modeled as 0-1 semidefinite optimization (0-1 SDO), a new optimization paradigm that covers many problems from various domains far beyond data clustering.   We also discuss how to utilize the 0-1 SDO model to develop new effective approaches for the underlying clustering problem. In the second part of this talk, we discuss how our study on the kernel feature selection problem in clustering motivated us to explore sparse solution to classes of quadratic optimization problems with random data , and how new sparse optimization models can help to cluster data sets with binary attributes. If time allows, we shall also discuss some future research directions.
Biography
Peng received PhD degree in operations research from Delft University of Technology, the Netherlands, 2001. Then he joined the department of computing and software in McMaster University, Ontario, Canada.  He joined the department of industrial and enterprise system engineering at University of Illinois at Urbana-Champaign in 2006, and moved to UH in the fall of 2013. His research interest lies mainly in optimization modeling, algorithm design and analysis, as well as application to data mining, financial engineering and health care.  So far he has published a research monograph by Princeton University, and over 50 papers in leading optimization journals such as Mathematical Programming, SIAM J. Optimization, Operations Research and top rank CS/IEEE conferences such as NIPS, CVPR etc. For his contributions in optimization, he received the Stieltjes Prize in 2001 and the Premier's research excellence award from Ontario in 2003.  He was also selected as one of the three finalists for the A.W Tucker prize (awarded triennially) from the mathematical optimization society in 2003. Very recently, one of his PhD students was selected as the first runner-up for the Annual Morgan Stanley Prize for Research Excellence in Financial Market.
Over the past years, Peng's research has been support by various funding agencies such as NSERC, MITACS and PREA in Canada, NSF (two awards) and AFOSR in USA. He has also organized numerous optimization meetings and has been serving as an associate editor for ``Optimization Letters'' since 2006.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://Bug.EGR.UH.EDU/pipermail/engi-dist/attachments/20131008/b0aefb86/attachment-0001.html