[Ml-stat-talks] Fwd: [talks] Colloquium Speaker Erik Sudderth Wed Nov 9- 4:30pm

David Blei blei at CS.Princeton.EDU
Sat Nov 5 09:49:24 EDT 2011

erik does great work and is a clear speaker.  this is not to be missed
if you are interested in graphical models, bayesian nonparametrics, or
computer vision.


---------- Forwarded message ----------
From: Nicole E. Wagenblast <nwagenbl at cs.princeton.edu>
Date: Fri, Nov 4, 2011 at 11:03 AM
Subject: [talks] Colloquium Speaker Erik Sudderth Wed Nov 9- 4:30pm
To: "Talks (colloquium)" <talks at lists.cs.princeton.edu>

Uncertainty in Natural Image Segmentation
Eric Sudderth, Brown University
Wednesday, November 9, 2011, 4:30 PM
Computer Science Small Auditorium (Room 105)

We explore nonparametric Bayesian statistical models for image
partitions which coherently model uncertainty in the size, shape, and
structure of human image interpretations. Examining a large set of
manually segmented scenes, we show that object frequencies and segment
sizes both follow power law distributions, which are well modeled by
the Pitman-Yor (PY) process. This generalization of the Dirichlet
process leads to segmentation algorithms which automatically adapt
their resolution to each image. Generalizing previous applications of
PY priors, we use non-Markov Gaussian processes (GPs) to infer
spatially contiguous segments which respect image boundaries. We show
how GP covariance functions can be calibrated to accurately match the
statistics of human segmentations, and that robust posterior inference
is possible via a variational method, expectation propagation. The
resulting method produces highly accurate segmentations of complex
scenes, and hypothesizes multiple image
 partitions to capture the variability inherent in human scene interpretations.
talks mailing list
talks at lists.cs.princeton.edu

More information about the Ml-stat-talks mailing list