Actions

Difference between revisions of "Events"

Computational Linguistics and Information Processing

(63 intermediate revisions by 8 users not shown)
Line 1: Line 1:
 
<center>[[Image:colloq.jpg|center|504px|x]]</center>
 
<center>[[Image:colloq.jpg|center|504px|x]]</center>
  
 +
== CLIP Colloquium ==
  
The CLIP Colloquium is a weekly speaker series organized and hosted by CLIP Lab. The talks are open to everyone. Most talks are held at 11AM in AV Williams 3258 unless otherwise noted. Typically, external speakers have slots for one-on-one meetings with Maryland researchers before and after the talks; contact the host if you'd like to have a meeting.
+
The CLIP Colloquium is a weekly speaker series organized and hosted by CLIP Lab. The talks are open to everyone. Most talks are held on Wednesday at 11AM in AV Williams 3258 unless otherwise noted. Typically, external speakers have slots for one-on-one meetings with Maryland researchers.
  
If you would like to get on the cl-colloquium@umiacs.umd.edu list or for other questions about the colloquium series, e-mail [mailto:jimmylin@umd.edu Jimmy Lin], the current organizer.
+
If you would like to get on the clip-talks@umiacs.umd.edu list or for other questions about the colloquium series, e-mail [mailto:oard@umiacs.umd.edu Doug Oard], the current organizer.
  
 +
For up-to-date information, see the [https://talks.cs.umd.edu/lists/7 UMD CS Talks page].  (You can also subscribe to the calendar there.)
  
{{#widget:Google Calendar
+
=== Colloquium Recordings ===
|id=lqah25nfftkqi2msv25trab8pk@group.calendar.google.com
+
* [[Colloqium Recording (Fall 2020)|Fall 2020]]
|color=B1440E
+
* [[Colloqium Recording (Spring 2021)|Spring 2021]]
|title=Upcoming Talks
 
|view=AGENDA
 
|height=300
 
}}
 
  
 +
=== Previous Talks ===
 +
* [[https://talks.cs.umd.edu/lists/7?range=past Past talks, 2013 - present]]
 +
* [[CLIP Colloquium (Spring 2012)|Spring 2012]]  [[CLIP Colloquium (Fall 2011)|Fall 2011]]  [[CLIP Colloquium (Spring 2011)|Spring 2011]]  [[CLIP Colloquium (Fall 2010)|Fall 2010]]
  
 +
== CLIP NEWS  ==
  
== 01/30/2013: Human Translation and Machine Translation ==
+
* News about CLIP researchers on the UMIACS website [http://www.umiacs.umd.edu/about-us/news]
 
+
* Please follow us on Twitter @umdclip [https://twitter.com/umdclip?lang=en]
'''Speaker:''' [http://homepages.inf.ed.ac.uk/pkoehn/ Philipp Koehn],  University of Edinburgh<br/>
 
'''Time:''' Wednesday, January 30, 2013, 11:00 AM<br/>
 
'''Venue:''' AVW 3258<br/>
 
 
 
Despite all the recent successes of machine translation, when it
 
comes to high quality publishable translation, human translators
 
are still unchallenged. Since we can't beat them, can we help
 
them to become more productive? I will talk about some recent
 
work on developing assistance tools for human translators.
 
You can also check out a prototype [http://www.caitra.org/ here]
 
and learn about our ongoing European projects [http://www.casmacat.eu/ CASMACAT]
 
and [http://www.matecat.com/ MATECAT].
 
 
 
'''About the Speaker:''' Philipp Koehn is Professor of Machine Translation at the
 
School of Informatics at the University of Edinburgh, Scotland.
 
He received his PhD at the University of Southern California
 
and spent a year as postdoctoral researcher at MIT.
 
He is well-known in the field of statistical machine translation
 
for the leading open source toolkit Moses, the organization
 
of the annual Workshop on Statistical Machine Translation
 
and its evaluation campaign as well as the Machine Translation
 
Marathon. He is founding president of the ACL SIG MT and
 
currently serves a vice president-elect of the ACL SIG DAT.
 
He has published over 80 papers and the textbook in the
 
field. He manages a number of EU and DARPA funded
 
research projects aimed at morpho-syntactic models, machine
 
learning methods and computer assisted translation tools.
 
 
 
== 02/06/2013: A New Recommender System for Large-scale Document Exploration ==
 
 
 
'''Speaker:''' [http://www.cs.cmu.edu/~chongw/ Chong Wang],  Carnegie Mellon University<br/>
 
'''Time:''' Wednesday, February 6, 2013, 11:00 AM<br/>
 
'''Venue:''' AVW 3258<br/>
 
 
 
How can we help people quickly navigate the vast amount of data
 
and acquire useful knowledge from it? Recommender systems provide
 
a promising solution to this problem. They narrow down the search
 
space by providing a few recommendations that are tailored to
 
users' personal preferences. However, these systems usually work
 
like a black box, limiting further opportunities to provide more
 
exploratory experiences to their users.
 
 
 
In this talk, I will describe how we build a new recommender
 
system for document exploration. Specially, I will talk about two
 
building blocks of the system in detail. The first is about a new
 
probabilistic model for document recommendation that is both
 
predictive and interpretable. It not only gives better predictive
 
performance, but also provides better transparency than
 
traditional approaches. This transparency creates many new
 
opportunities for exploratory analysis---For example, a user can
 
manually adjust her preferences and the system responds to this
 
by changing its recommendations. Second, building a recommender
 
system like this requires learning the probabilistic model from
 
large-scale empirical data. I will describe a scalable approach
 
for learning a wide class of probabilistic models that include
 
our recommendation model as a special case.
 
 
 
'''About the Speaker:''' Chong is a Project Scientist in Eric Xing's group, Machine Learning Department, Carnegie Mellon University.  His PhD advisor was David M. Blei from Princeton University.
 
 
 
== 02/13/2013: Mona Diab ==
 
 
 
'''Speaker:''' [http://www1.ccls.columbia.edu/~mdiab/ Mona Diab], Columbia University<br/>
 
'''Time:''' Wednesday, February 13, 2013, 11:00 AM<br/>
 
'''Venue:''' AVW 3258<br/>
 
 
 
== 02/14/2013: Efficient Probabilistic Models for Rankings and Orderings ==
 
 
 
'''Speaker:''' [http://stanford.edu/~jhuang11/ Jon Huang], Stanford University<br/>
 
'''Time:''' Thursday, February 14, 2013, 11:00 AM<br/>
 
'''Venue:''' AVW 3258<br/>
 
 
 
The need to reason probabilistically with rankings and orderings arises
 
in a number of real world problems.  Probability distributions over
 
rankings and orderings arise naturally, for example, in preference data,
 
and political election data, as well as a number of less obvious
 
settings such as topic analysis and neurodegenerative disease
 
progression modeling. Representing distributions over the space of all
 
rankings is challenging, however, due to the factorial number of ways to
 
rank a collection of items.  The focus of my talk is to discuss methods
 
for combatting this factorial explosion in probabilistic representation
 
and inference.
 
 
 
Ordinarily, a typical machine learning method for dealing with
 
combinatorial complexity might be to exploit conditional independence
 
relations in order to decompose a distribution into compact factors of a
 
graphical model.  For ranked data, however, a far more natural and
 
useful probabilistic relation is that of `riffled independence'.  I will
 
introduce the concept of riffled independence and discuss how these
 
riffle independent relations can be used to decompose a distribution
 
over rankings into a product of compactly represented factors.  These
 
so-called hierarchical riffle-independent distributions are particularly
 
amenable to efficient inference and learning algorithms and in many
 
cases lead to intuitively interpretable probabilistic models. To
 
illustrate the power of exploiting riffled independence, I will discuss
 
a few applications, including Irish political election analysis,
 
visualizing the japanese preferences of sushi types and modeling the
 
progression of Alzheimer's disease, showing results on real datasets in
 
each problem.
 
 
 
This is joint work with Carlos Guestrin (University of Washington),
 
Ashish Kapoor (Microsoft Research) and Daniel Alexander (University
 
College London).
 
 
 
== 02/27/2013:David Mimno ==
 
 
 
== 03/13/2013: Dan Hopkins ==
 
 
 
== 03/27/2013: Richard Sproat ==
 
 
 
== 04/10/2013: Learning with Marginalized Corrupted Features ==
 
 
 
'''Speaker:''' [http://www.cse.wustl.edu/~kilian/ Kilian Weinberger],  Washington University in St. Louis<br/>
 
'''Time:''' Wednesday, April 10, 2013, 11:00 AM<br/>
 
'''Venue:''' AVW 3258<br/>
 
 
 
If infinite amounts of labeled data are provided, many machine learning algorithms become perfect. With finite amounts of data, regularization or priors have to be used to introduce bias into a classifier. We propose a third option: learning with marginalized corrupted features. We corrupt existing data as a means to generate infinitely many additional training samples from a slightly different data distribution -- explicitly in a way that the corruption can be marginalized out in closed form. This leads to machine learning algorithms that are fast, effective and naturally scale to very large data sets. We showcase this technology in two settings: 1. to learn text document representations from unlabeled data and 2. to perform supervised learning with closed form gradient updates for empirical risk minimization.
 
 
 
Text documents (and often images) are traditionally expressed as bag-of-words feature vectors (e.g. as tf-idf). By training linear denoisers that recover unlabeled data from partial corruption, we can learn new data-specific representations. With these, we can match the world-record accuracy on the Amazon transfer learning benchmark with a simple linear classifier. In comparison with the record holder (stacked denoising autoencoders) our approach shrinks the training time from several days to a few minutes.
 
 
 
Finally, we present a variety of loss functions and corrupting distributions, which can be applied out-of-the-box with empirical risk minimization. We show that our formulation leads to significant improvements in document classification tasks over the typically used l_p norm regularization. The new learning framework is extremely versatile, generalizes better, is more stable during test-time (towards distribution drift) and only adds a few lines of code to typical risk minimization. 
 
 
 
'''About the Speaker:''' Kilian Q. Weinberger is an Assistant Professor in the Department of Computer Science & Engineering at Washington University in St. Louis. He received his Ph.D. from the University of Pennsylvania in Machine Learning under the supervision of Lawrence Saul. Prior to this, he obtained his undergraduate degree in Mathematics and Computer Science at the University of Oxford. During his career he has won several best paper awards at ICML, CVPR and AISTATS. In 2011 he was awarded the AAAI senior program chair award and in 2012 he received the NSF CAREER award. Kilian Weinberger's research is in Machine Learning and its applications. In particular, he focuses on high dimensional data analysis, metric learning, machine learned web-search ranking, transfer- and multi-task learning as well as bio medical applications.
 
 
 
 
 
== Previous Talks ==
 
* [[CLIP Colloquium (Fall 2012)|Fall 2012]]
 
* [[CLIP Colloquium (Spring 2012)|Spring 2012]]
 
* [[CLIP Colloquium (Fall 2011)|Fall 2011]]
 
* [[CLIP Colloquium (Spring 2011)|Spring 2011]]
 
* [[CLIP Colloquium (Fall 2010)|Fall 2010]]
 

Revision as of 18:21, 6 June 2021

x

CLIP Colloquium

The CLIP Colloquium is a weekly speaker series organized and hosted by CLIP Lab. The talks are open to everyone. Most talks are held on Wednesday at 11AM in AV Williams 3258 unless otherwise noted. Typically, external speakers have slots for one-on-one meetings with Maryland researchers.

If you would like to get on the clip-talks@umiacs.umd.edu list or for other questions about the colloquium series, e-mail Doug Oard, the current organizer.

For up-to-date information, see the UMD CS Talks page. (You can also subscribe to the calendar there.)

Colloquium Recordings

Previous Talks

CLIP NEWS

  • News about CLIP researchers on the UMIACS website [1]
  • Please follow us on Twitter @umdclip [2]