Unimore logo AImageLab

Constraint Propagation and Value Acquisition: why we should do it Interactively

Abstract: In Constraint Satisfaction Problems (CSPs) values belonging to variable domains should be completely known before the constraint propagationprocess starts. In many applications, however, the acquisition of domain values is a computational expensive process or some domainvalues could not be available at the beginningof the computation. For this purpose, we introduce an Interactive Constraint SatisfactionProblem (ICSP) model as extension of the widely used CSP model. The variable domainvalues can be acquired when needed duringthe resolution process by means of InteractiveConstraints, which retrieve (possibly consistent)information. Experimental results on randomly generated CSPs and for 3D object recognition show the effectiveness of the proposedapproach.


Citation:

E., Lamma; P., Mello; M., Milano; Cucchiara, Rita; G., Gavanelli; M., Piccardi "Constraint Propagation and Value Acquisition: why we should do it Interactively" Proceedings of the Sixteenth International Joint Conference on Artificial Intelligence, Stockholm, Sweden, pp. 468 -477 , July 31 - Aug. 6, 1999

 not available

Paper download:

  • Author version: