BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IC Colloquium : Efficient inference with combinatorial structure
DTSTART:20130411T150000
DTEND:20130411T163000
DTSTAMP:20260406T172814Z
UID:6f1f62dae8ad08531cf4c4ccd87e6391341abe04b658db53cc735a8a
CATEGORIES:Conferences - Seminars
DESCRIPTION:Stefanie Jegelka\, University of California Berkeley\nIC facul
 ty candidate\nAbstract\nLearning from complex data such as images or biolo
 gical measurements invariably relies on capturing latent structure. But th
 e combinatorial structure inherent in real-world data poses significant co
 mputational challenges for modeling\, learning and inference\, bringing co
 mmonly used probabilistic graphical models to their limits. In this talk\,
  I will describe work towards overcoming these challenges. As a concrete e
 xample\, I will introduce a novel\, rich class of models whose inference p
 rocedures are not only efficient but also enjoy provable approximation bou
 nds. This class of models combines two versatile concepts: graphs and subm
 odular functions. While in its general form\, this class can lead to hard 
 inference problems\, several practically relevant properties admit improve
 d approximations or even fast exact inference. These results enable the re
 alization of new models\, such as a combinatorial sparsity that significan
 tly improves image segmentation results in settings where state-of-the art
  methods fail. I will use this illustrative example to outline theoretical
  and empirical results\, and will place them in a broader context within a
 nd beyond machine learning\, sketching relations to structured sparsity\, 
 hierarchical models and implications for recently studied submodular optim
 ization problems.Biography\nStefanie Jegelka is a postdoctoral researcher 
 at UC Berkeley\, supervised by Michael Jordan and Trevor Darrell. She rece
 ived a Ph.D. in Computer Science from ETH Zurich in 2012\, in collaboratio
 n with the Max Planck Institute for Intelligent Systems and under the supe
 rvision of Jeff Bilmes and Bernhard Schölkopf. She completed her studies 
 for a Diplom in Bioinformatics with distinction at the University of Tuebi
 ngen and the University of Texas at Austin. During her Diplom studies\, sh
 e was a fellow of the German National Academic Foundation (Studienstiftung
 ) and a member of its scientific college for life sciences. She also recei
 ved a Google Anita Borg Europe Fellowship and has been a research visitor 
 at Georgetown University Medical Center and at Microsoft Research. Her res
 earch interests lie in algorithms and machine learning.
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR