Olgica Milenkovic
Electrical and Computer Engineering
University of Illinois
Faculty Host: Prakash Ishwar
Light refreshments will be served outside of PHO 428 at 1:45 pm.
New Directions in Correlation Clustering and Biclustering inspired by Bioinformatics and Social Sciences
Abstract:
In the most basic form of the correlation clustering (CC) model, one is given a set of objects and, for some pairs of objects, one is also given an assessment as to whether the objects are “similar” or “dissimilar”. This information is described using a graph G with labeled edges: each object is represented by a vertex of the graph, and the assessments are represented by edges labeled with either a + (for similar objects) or a − (for dissimilar objects). The goal is to partition the objects into clusters so that the edges within clusters are mostly positive and the edges between clusters are mostly negative. Unlike many other clustering models, such as k-means, the number of clusters is not fixed ahead of time. Instead, finding the optimal number of clusters is part of the problem.
Although many different modifications of the CC model were studied in the past, most of them ignore practical constraints arising from emerging high volume data applications in bioinformatics and social sciences, where one requires highly specific clustering constraints and individual quality of service criteria for each vertex in the graphs. We describe a number of such paradigms and highlight their application in cancer genomics, recommender systems and data deduplication.
A Joint Work with Amin Emad, Jian Ma, and Gregory Puleo, UIUC
Bio:
Olgica Milenkovic is a professor of Electrical and Computer Engineering at the University of Illinois, Urbana-Champaign (UIUC), and Research Professor at the Coordinated Science Laboratory. She obtained her Masters Degree in Mathematics in 2001 and PhD in Electrical Engineering in 2002, both from the University of Michigan, Ann Arbor. From 2002 to 2006, she was with the University of Colorado, Boulder, and in 2005 she also served as consultant for the Bell Laboratories. During the years 2006 and 2007 she visited the University of California, San Diego, and then subsequently joined the University of Illinois in 2007. She is a member of the NSF Center for Science of Information and was a long term visitor of the Simons Institute at Berkeley as part of the 2015 Information Theory program. Prof. Milenkovic heads a group focused on addressing unique interdisciplinary research challenges spanning the areas of algorithm design and computing, bioinformatics, coding theory, machine learning and signal processing.
|
When |
Friday, Aug 28, 2015
at 2:00pm
until 3:00pm
on Friday, Aug 28, 2015
|