Skip to Content.
Sympa Menu

nl-uiuc - [nl-uiuc] Reminder: AIIS talk: Roni Khardon ( _Today_ )

nl-uiuc AT lists.cs.illinois.edu

Subject: Natural language research announcements

List archive

[nl-uiuc] Reminder: AIIS talk: Roni Khardon ( _Today_ )


Chronological Thread 
  • From: Yonatan Bisk <bisk1 AT illinois.edu>
  • To: nl-uiuc AT cs.uiuc.edu, aivr AT cs.uiuc.edu, dais AT cs.uiuc.edu, cogcomp AT cs.uiuc.edu, vision AT cs.uiuc.edu, eyal AT cs.uiuc.edu, aiis AT cs.uiuc.edu, aistudents AT cs.uiuc.edu, "Girju, Corina R" <girju AT illinois.edu>, Catherine Blake <clblake AT illinois.edu>, "Efron, Miles James" <mefron AT illinois.edu>
  • Subject: [nl-uiuc] Reminder: AIIS talk: Roni Khardon ( _Today_ )
  • Date: Fri, 4 Mar 2011 11:03:46 -0600
  • List-archive: <http://lists.cs.uiuc.edu/pipermail/nl-uiuc>
  • List-id: Natural language research announcements <nl-uiuc.cs.uiuc.edu>

When:       _Today_  @ 2pm

Where:      3405 SC

Speaker:   Roni Khardon  ( http://www.cs.tufts.edu/~roni/ )

Title:          First Order Decision Diagrams for Relational Markov Decision Processes

Abstract:   
Markov decision processes (MDP) capture sequential decision making under uncertainty, where an agent must choose actions so as to optimize long term reward. Relational MDPs capture problems where world states have an internal relational structure that can be naturally described in terms of objects and relations among them. For example, a system managing shipment of packages using trucks and airplanes must optimize for short delivery time and low costs. Such problems are often huge in size and yet offer potential for efficient algorithmic solutions taking advantage of their structure. 

The talk will first introduce First Order Decision Diagrams (FODD), a new compact representation for functions over relational structures, generalizing the well known binary decision diagrams. It will then show how a "calculus of FODDs" can be used to solve relational MDPs at the abstract level using a "lifted" form of the value iteration algorithm. Several extensions of this approach will be briefly discussed handling more expressive representations, lifting other algorithms, and coping with partial observability. Finally results of a prototype implementation will demonstrate the potential of this approach in solving challenging problems.


Bio:        Associate Professor of Computer Science at Tufts University.

- Yonatan -



  • [nl-uiuc] Reminder: AIIS talk: Roni Khardon ( _Today_ ), Yonatan Bisk, 03/04/2011

Archive powered by MHonArc 2.6.16.

Top of Page