The Knowledge Engineering Review
ATC'10 Proceedings of the 7th international conference on Autonomic and trusted computing
LEARNING AND VERIFYING SAFETY CONSTRAINTS FOR PLANNERS IN A KNOWLEDGE-IMPOVERISHED SYSTEM
Computational Intelligence
| Hi-index | 0.00 |
Constraint programming offers a declarative approach to solving problems modeled as constraint satisfaction prob- lems (CSPs). However, the precise specification of a set of constraints is sometimes not available, but may have to be learned, for instance, from a set of examples of its solutions and non-solutions. In general, one may wish to learn gen- eralized CSPs involving classical, fuzzy, weighted or prob- abilistic constraints, for example. This paper introduces a unifying framework for CSP learning. The framework is generic in that it can be instantiated to obtain specific for- mulations for learning classical, fuzzy, weighted or prob- abilistic CSPs. In particular, a new formulation for clas- sical CSP learning, which minimizes the number of exam- ples violated by candidate CSPs, is obtained by instantiat- ing the framework. This formulation is equivalent to a sim- ple pseudo-boolean optimization problem, thus being effi- ciently solvable using many optimization tools.