Feature allocations, probability functions, and paintboxes

Speaker: Tamara Broderick , University of California - Berkeley

Date: Monday, March 10, 2014

Time: 4:00 PM to 5:00 PM Note: all times are in the Eastern Time Zone

Refreshments: 3:45 PM

Public: Yes

Location: 32-G449

Event Type:

Room Description:

Host: Regina Barzilay and Tommi Jaakkola, CSAIL

Contact: Francis Doughty, 253-4602, francisd@csail.mit.edu

Relevant URL:

Speaker URL: None

Speaker Photo:
None

Reminders to: seminars@csail.mit.edu

Reminder Subject: TALK: Tamara Broderick: Feature allocations, probability functions, and paintboxes

Abstract:
Clustering involves placing entities into mutually exclusive
categories.  We wish to relax the requirement of mutual exclusivity,
allowing objects to belong simultaneously to multiple classes, a
formulation that we refer to as "feature allocation."  The first step
is a theoretical one.  In the case of clustering the class of
probability distributions over exchangeable partitions of a dataset
has been characterized (via exchangeable partition probability
functions and the Kingman paintbox).  These characterizations support
an elegant nonparametric Bayesian framework for clustering in which
the number of clusters is not assumed to be known a priori.  We
establish an analogous characterization for feature allocation; we
define notions of "exchangeable feature probability functions" and
"feature paintboxes" that lead to a Bayesian framework that does not
require the number of features to be fixed a priori.  The second step
is a computational one.  Rather than appealing to Markov chain Monte
Carlo for Bayesian inference, we develop a method to transform
Bayesian methods for feature allocation (and other latent structure
problems) into optimization problems with objective functions
analogous to K-means in the clustering setting.  These yield
approximations to Bayesian inference that are scalable to large
inference problems.

Bio:
Tamara Broderick is a PhD candidate in the Department of Statistics at
the University of California, Berkeley. Her research in machine
learning focuses on the design and study of Bayesian nonparametric
models, with particular emphasis on feature allocation as a
generalization of clustering that relaxes the mutual exclusivity and
exhaustivity assumptions of clustering. While at Berkeley, she has
been a National Science Foundation Graduate Student Fellow and a
Berkeley Fellowship recipient. She graduated with an AB in Mathematics
from Princeton University in 2007---with the Phi Beta Kappa Prize for
highest average GPA in her graduating class and with Highest Honors in
Mathematics. She spent the next two years on a Marshall Scholarship at
the University of Cambridge, where she received a Masters of Advanced
Study in Mathematics for completion of Part III of the Mathematical
Tripos (with Distinction) in 2008 and an MPhil by Research in Physics
in 2009. She received a Masters in Computer Science from UC Berkeley
in 2013.

Research Areas:

Impact Areas:

See other events that are part of the CS Special Seminar Series 2014.

Created by Francis Doughty at Thursday, February 06, 2014 at 1:37 PM.