<div dir="ltr">Reminder... Victor's PhD defense starts in 20 minutes, in the CS conference room.<div><br></div><div> Matthew</div><div><br><div class="gmail_quote">---------- Forwarded message ----------<br>From: <b class="gmail_sendername">Jillian Title</b> <span dir="ltr"><<a href="mailto:jillian.title@cs.ucsb.edu" target="_blank">jillian.title@cs.ucsb.edu</a>></span><br>Date: Mon, Dec 1, 2014 at 4:30 PM<br>Subject: [FACULTY] PhD Defense - Victor Fragoso - 12/5/14<br>To: <a href="mailto:faculty@cs.ucsb.edu" target="_blank">faculty@cs.ucsb.edu</a>, <a href="mailto:grads@cs.ucsb.edu" target="_blank">grads@cs.ucsb.edu</a>, <a href="mailto:lecturers@cs.ucsb.edu" target="_blank">lecturers@cs.ucsb.edu</a>, <a href="mailto:research@lists.cs.ucsb.edu" target="_blank">research@lists.cs.ucsb.edu</a><br><br><br>
<div bgcolor="#FFFFFF" text="#000000">
PhD Defense<br>
<b>Victor Fragoso</b><br>
Monday, December 8th at 2pm<br>
HFH 1132<br>
<br>
<b>Committee:</b> Matthew Turk (Chair), Joao Hespanha, Pradeep Sen,
Walter Scheirer<br>
<b><br>
</b><b>Title: </b>Estimating Confidences for Classifier’s Decisions
using Extreme Value Theory<br>
<br>
<b>Abstract: </b><br>
<br>
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.<br>
<br>
Everyone Welcome!<span><font color="#888888"><br>
<br>
<pre cols="72">--
Jillian Title
Graduate Advisor
Department of Computer Science
University of California Santa Barbara
2104 Harold Frank Hall
Santa Barbara, CA 93106-5110
<a href="tel:%28805%29%20893-4322" value="+18058934322" target="_blank">(805) 893-4322</a></pre>
</font></span></div>
<br>_______________________________________________<br>
faculty mailing list<br>
<a href="mailto:faculty@lists.cs.ucsb.edu" target="_blank">faculty@lists.cs.ucsb.edu</a><br>
<a href="https://lists.cs.ucsb.edu/mailman/listinfo/faculty" target="_blank">https://lists.cs.ucsb.edu/mailman/listinfo/faculty</a><br>
<br></div><br></div></div>