[4eyes] Fwd: [FACULTY] PhD Defense - Victor Fragoso - 12/5/14

[Matthew Turk]

Reminder... Victor's PhD defense starts in 20 minutes, in the CS conference
room.

    Matthew

---------- Forwarded message ----------
From: Jillian Title <jillian.title at cs.ucsb.edu>
Date: Mon, Dec 1, 2014 at 4:30 PM
Subject: [FACULTY] PhD Defense - Victor Fragoso - 12/5/14
To: faculty at cs.ucsb.edu, grads at cs.ucsb.edu, lecturers at cs.ucsb.edu,
research at lists.cs.ucsb.edu


 PhD Defense
*Victor Fragoso*
Monday, December 8th at 2pm
HFH 1132

*Committee:* Matthew Turk (Chair), Joao Hespanha, Pradeep Sen, Walter
Scheirer

*Title: *Estimating Confidences for Classifier’s Decisions using Extreme
Value Theory

*Abstract: *

Classifiers generally lack a mechanism to compute decision confidences. As
humans, when we sense that the confidence for a decision is low, we either
conduct additional actions to improve our confidence or dismiss the
decision. While this reasoning is natural to us, it is currently missing in
most current decision algorithms (i.e., classifiers) used in computer
vision or machine learning. This limits the capability for a machine to
take further actions to either improve a result or dismiss the decision. In
this thesis, we design algorithms for estimating the confidence for
decisions made by classifiers, such as, nearest-neighbor, or support vector
machines. We developed these algorithms leveraging the theory of extreme
values. We use the statistical models that this theory provides for
modeling the classifier's decision scores of correct and incorrect
outcomes. Our proposed algorithms exploit these statistical models in order
to compute a correctness belief: the probability that the classifier's
decision is correct. We show how these beliefs can be used to filter bad
classifications; and to speed up robust estimations via sample and
consensus algorithms, which are used in computer vision for estimating
camera motions and for reconstructing the scene's 3D structure. Moreover,
we show how these beliefs improve the classification accuracy of one-class
support vector machines. In conclusion, we show that extreme value theory
leads to powerful mechanisms that can predict the correctness of a
classifier's decision.

Everyone Welcome!

-- 
Jillian Title
Graduate Advisor
Department of Computer Science
University of California Santa Barbara
2104 Harold Frank Hall
Santa Barbara, CA 93106-5110(805) 893-4322


_______________________________________________
faculty mailing list
faculty at lists.cs.ucsb.edu
https://lists.cs.ucsb.edu/mailman/listinfo/faculty
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cs.ucsb.edu/pipermail/ilab-users/attachments/20141208/a1eee259/attachment.html>