Applied partial constraint satisfaction using weighted iterative repair
Many real-world constraint satisfaction problems (CSPs) can be over-constrained or too large to solve using a standard constructive/ backtracking approach. Instead, faster heuristic techniques have been proposed that perform a partial search of all possible solutions using an iterative repair or hill-climbing approach. The main problem with such approaches is that they can become stuck in local minima. Consequently, various strategies or metaheuristics have been developed to escape from local minima. This paper investigates the application of one such meta-heuristic, weighted iterative repair, to solving a real-world problem of scheduling nurses at an Australian hospital. Weighted iterative repair has already proved successful in solving various binary CSPs. The current research extends this work by looking at a non-binary problem formulation, and partial constraint satisfaction involving hard and soft constraints. This has lead to the development of a soft constraint heuristic to improve the level of soft constraint optimisation and an extension of the original weighted iterative repair that avoids certain forms of cyclic behaviour. It is also demonstrated that weighted iterative repair can learn from repeatedly solving the same problem. and that restarting the algorithm on the same problem can result in faster execution times. The overall results show that weighted iterative repair finds better quality solutions than a standard iterative repair, whilst approaching near optimal solutions in less time than an alternative integer programming approach.
KeywordsConstraint Satisfaction Constraint Satisfaction Problem Soft Constraint Hard Constraint Constraint Weight
Unable to display preview. Download preview PDF.
- 1.D. Abramson. A very high speed architecture for simulated annealing. IEEE Comp., May:27–36, 1992.Google Scholar
- 2.A. Borning, B. Freeman-Benson and M. Wilson. Constraint hierarchies. In M. Jampel, E. Freuder and M. Maher, editors, Over-Constrained Systems, pages 23–62. Springer-Verlag, 1996.Google Scholar
- 3.B. Cheng, J. Lee and J. Wu. A constraint-based nurse rostering system using a redundant modeling approach. In Proc. of the 8th IEEE International Conference on Tools with AI, 1996.Google Scholar
- 4.M. S. Fox. ISIS: A Retrospective. In M. Zweben and M. S. Fox, editors, Intelligent Scheduling, pages 3–28. Morgan Kaufman, 1994.Google Scholar
- 6.F. Glover. Tabu search — part 1. ORSA Journal on Computing, 1(3):190–206, 1989.Google Scholar
- 7.H. E. Miller, W. P. Pierskalla and G. J. Rath. Nurse scheduling using mathematical programming. Ops. Res., 24(5):857–870, 1976.Google Scholar
- 9.P. Morris. The breakout method for escaping local from minima. In Proc. of AAAI'93, pages 40–45, 1993.Google Scholar
- 10.B. Selman and H. Kautz. Domain independent extensions to GSAT: Solving large structured satisfiability problems. In Proc. of IJCAI 93, pages 290–295, 1993.Google Scholar
- 11.J. R. Thornton and A. Sattar. An integer programming-based nurse rostering system. In Proc. ofASIAN `96, pages 357–358, Singapore, 1996.Google Scholar
- 12.R. J. Wallace and E. C. Freuder. Heuristic methods for over-constrained constraint satisfaction problems. In M. Jampel, E. Freuder and M. Maher, editors, Over-Constrained Systems, pages 207–216. Springer-Verlag, 1996.Google Scholar
- 13.D. M. Warner. Scheduling nursing personnel according to nursing preference: A mathematical programming approach. Ops. Res. 24(5):842–856, 1976.Google Scholar