BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IC Colloquium : Scalable Inference and Learning for High-Level Pro
 babilistic Models
DTSTART:20150226T101500
DTEND:20150226T113000
DTSTAMP:20260407T163622Z
UID:ef96a6c8f7560e46094d999ea09c0f11806a87e7ac3779e345bc6062
CATEGORIES:Conferences - Seminars
DESCRIPTION:By : Guy Van den Broeck - KU Leuven\nIC Faculty candidateAbstr
 act :\nProbabilistic graphical models are pervasive in AI and machine lear
 ning. A recent push\, however\, is towards more high-level representations
  of uncertainty\, such as probabilistic programs\, probabilistic databases
 \, and statistical relational models. This move is akin to going from hard
 ware circuits to a full-fledged programming language\, and poses key chall
 enges for inference and learning. For instance\, we encounter a fundamenta
 l limitation of classical learning algorithms: they make strong independen
 ce assumptions about the entities in the data (e.g.\, images\, web pages\,
  etc.). These assumptions fail to hold in a global view of the data\, wher
 e all entities are related. We also encounter a limitation of existing rea
 soning algorithms\, which fail to scale to large\, densely connected graph
 ical models\, consisting of millions of interrelated entities.\nIn this ta
 lk\, I present my research on efficient algorithms for high-level probabil
 istic models\, called lifted inference and learning algorithms. I begin by
  introducing the key principles behind exact lifted inference\, namely to 
 exploit symmetry and exchangeability in the model. Next\, I discuss the st
 rengths and limitations of lifting. Building on results from database theo
 ry and counting complexity\, I identify classes of tractable models\, and 
 classes where high-level reasoning is fundamentally hard. I conclude by sh
 owing the practical embodiment of these ideas\, in the form of approximate
  inference and learning algorithms that scale up to big data and big model
 s.Bio :\nGuy Van den Broeck graduated summa cum laude with a Ph.D. in Comp
 uter Science from KU Leuven\, Belgium\, in 2013. He was a postdoctoral res
 earcher at UCLA and KU Leuven. His research interests are broadly in machi
 ne learning\, artificial intelligence\, knowledge representation and reaso
 ning\, and statistical relational learning. His work was awarded the ECCAI
  AI Dissertation Award 2014\, Scientific prize IBM Belgium for Informatics
  2014\, and Alcatel-Lucent Innovation Award 2009. He is the recipient of t
 he best student paper award at ILP 2011 and a best paper honorable mention
  at AAAI 2014.More information
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR