Bucket Elimination: a Unifying Framework for ProcessingHard and Soft Constraints

  • Authors:
  • Rina Dechter

  • Affiliations:
  • Department of Information and Computer Science, University of California, Irvine, Irvine, CA 92697-3425. e-mail: dechter@ics.uci.edu

  • Venue:
  • Constraints
  • Year:
  • 1997

Quantified Score

Hi-index 0.00

Visualization

Abstract

This position paper argues that extending the CSP modelto a richer set of tasks such as, constraint optimization, probabilisticinference and decision theoretic tasks can be done within a unifyingframework called ’’bucket elimination‘‘. The framework allowsuniform hybrids for combining elimination and conditioning guidedby the problem‘s structure and for explicating the tradeoffsbetween space and time and between time and accuracy.