[Compstats] AI/Compstats seminar Friday @ 11:30 (DC2306C): Rita Ackerman

Pascal Poupart ppoupart at cs.uwaterloo.ca
Mon May 3 18:36:22 EDT 2010


Hi everyone,

In addition to David McAllester's guest lecture in the Google-UW seminar 
series on Wednesday, Rita Ackerman will give a Comptstats/AI seminar on 
Friday.  Title and abstract follow below.

cheers, Pascal
===========

Title:  Characterization of Linkage-Based Clustering

Speaker: Rita Ackerman
Time: May 7, 2010 at 11:30 am
Location: AI Lab (DC 2306C)

Abstract:
Clustering is a central unsupervised learning task with a wide varietyof 
applications. Not surprisingly,
there exist many clustering algorithms. However, unlike classification 
tasks, in clustering,  different algorithms may yield dramatically 
different outputs for the same input sets. A major challenge is to 
develop tools that may help select the more suitable algorithm for a 
given clustering task. We propose to address this problem by distilling 
abstract properties of clustering functions that distinguish between the 
types of input-output behaviors of different clustering paradigms. In 
this talk we make a step in this direction by providing such property 
based characterization for the class of linkage based clustering 
algorithms.

Linkage-based clustering is one the most commonly used and widely 
studied clustering paradigms. It includes popular algorithms like Single 
Linkage and enjoys simple efficient algorithms. On top of their 
potential merits for helping users decide when are such algorithms 
appropriate
for their data, our results can be viewed as a convincing proof of 
concept for the research on taxonomizing clustering paradigms by their 
abstract properties.

-- 
------------------------
Pascal Poupart
Associate Professor
David R. Cheriton School of Computer Science
University of Waterloo
200 University Avenue West
Waterloo, Ontario
Canada N2L 3G1
------------------------
Web: http://www.cs.uwaterloo.ca/~ppoupart
Email: ppoupart at cs.uwaterloo.ca 
Telephone: 1-519-888-4567x36239 
Fax: 1-519-885-1208
------------------------



More information about the Compstats mailing list