AIC Seminar Series
Multi-Modal Clustering: a Formal Framework and Practical Applications
|Ron Bekkerman||University of Massachusetts Amherst|
Notice: hosted by Jeffrey Davitz
Date: Monday August 06, 2007 at 10:00
Location: EJ228 (SRI E building) (Directions)
In this talk, after giving a brief overview of my research, I will focus
on one of my favorite topics in machine learning: multi-modal clustering,
which is a clustering problem in the environment where multiple views (or
modalities) of the input data are available. In text clustering, for
instance, modalities are documents, their words, their authors, titles,
markup primitives etc. Multi-modal clustering is a problem of
simultaneously constructing N partitionings of N data modalities, which
reduces statistical sparseness of data representation, and potentially
leads to more accurate clusterings than those obtained separately.
First, I will present Combinatorial Markov Random Fields (Comrafs) that I
have recently proposed as a formal framework for multi-modal clustering.
Comrafs have proved themselves to be the current state-of-the-art in
document clustering. Second, I will discuss particular modeling choices
that result in deriving a variety of Comraf models not only for basic text
clustering, but also for image clustering, semi-supervised and interactive
clustering, and other tasks. Finally, I will address the efficiency issue
and propose a Comraf model able to accurately cluster a multi-million
Ron Bekkerman is a PhD candidate at the University of Massachusetts
Amherst, working under the supervision of Prof. James Allan on
unsupervised and semi-supervised learning problems, with applications to
information retrieval and Web mining. Ron received his BSc and MSc in CS
from the TechnionIsrael Institute of Technology. His Masters thesis
was on feature induction for text categorization.
Please arrive at least 10 minutes early in order to sign in and be escorted to the conference room. SRI is located at 333 Ravenswood Avenue in Menlo Park. Visitors may park in the visitors lot in front of Building E, and should follow the instructions by the lobby phone to be escorted to the meeting room. Detailed directions to SRI, as well as maps, are available from the Visiting AIC web page.
©2013 SRI International 333 Ravenswood Avenue, Menlo Park, CA 94025-3493