Title: Task Scheduling using Constraint Optimization with Uncertainty
Speaker: James Atlas
Department of Computer & Information Sciences, University of Delaware
Abstract:
This presentation will be mostly on unpublished research towards the
application portion of my dissertation. The talk title is actually
taken from a short workshop paper I had published last year at
AAMAS2007 which contained some initial formulations and direction for
my current research. Here is the abstract for the talk:
Multiagent task scheduling encompasses diverse domains of problems
that require complex models and robust solutions. C-TÆMS is a new
specification, developed for use by DARPA's COORDINATORs initiative
and based on TÆMS, for multiagent task scheduling problems that
represents the complex relationships necessary to model these domains.
Multiagent task scheduling has many applications including automated
project management, real-time team coordination, and execution of grid
and web-service workflows.
In this talk I will present an extended version of distributed
constraint optimization which handles uncertainty in problem
parameters. In particular, I will apply this framework to C-TÆMS
problems used in the phase competitions of the COORDINATORs
initiative. I will present initial results and outline the strengths
and weaknesses of the approach.
Special note for attendees:
This research is part of my
dissertation work I am especially interested in feedback. Most of the
presented work will be unpublished, so please speak with me if you
would like to discuss the work with anyone outside of our department.