Handling Duration Uncertainty in Meta-Level Control of Progressive Processing

Abdel-Illah Mouaddib and Shlomo Zilberstein. Handling Duration Uncertainty in Meta-Level Control of Progressive Processing. Proceedings of the Fifteenth International Joint Conference on Artificial Intelligence (IJCAI), 1201-1207, Nagoya, Japan, 1997.

Abstract

Progressive processing is a resource-bounded reasoning technique that allows a system to incrementally construct a solution to a problem using a hierarchy of processing levels. This paper focuses on the problem of meta-level control of progressive processing in domains characterized by rapid change and high level of duration uncertainty. We show that progressive processing facilitates efficient run-time monitoring and meta-level control. Our solution is based on an incremental scheduler that can handle duration uncertainty by dynamically revising the schedule during execution time based on run-time information. We also show that a probabilistic representation of duration uncertainty reduces the frequency of schedule revisions and thus improves the performance of the system. Finally, an experimental evaluation shows the contributions of this approach and its suitability for a data transmission application.

Bibtex entry:

@inproceedings{MZijcai97,
  author	= {Abdel-Illah Mouaddib and Shlomo Zilberstein},
  title		= {Handling Duration Uncertainty in Meta-Level Control of
                   Progressive Processing},
  booktitle     = {Proceedings of the Fifteenth International Joint Conference
                   on Artificial Intelligence},
  year		= {1997},
  pages		= {1201-1207},
  address       = {Nagoya, Japan},
  url		= {http://rbr.cs.umass.edu/shlomo/papers/MZijcai97.html}
}

shlomo@cs.umass.edu
UMass Amherst