A Racing Algorithm for Configuring Metaheuristics
GECCO '02 Proceedings of the Genetic and Evolutionary Computation Conference
ICML '09 Proceedings of the 26th Annual International Conference on Machine Learning
ParamILS: an automatic algorithm configuration framework
Journal of Artificial Intelligence Research
Improvement strategies for the F-Race algorithm: sampling design and iterative refinement
HM'07 Proceedings of the 4th international conference on Hybrid metaheuristics
Connections in networks: a hybrid approach
CPAIOR'08 Proceedings of the 5th international conference on Integration of AI and OR techniques in constraint programming for combinatorial optimization problems
A gender-based genetic algorithm for the automatic configuration of algorithms
CP'09 Proceedings of the 15th international conference on Principles and practice of constraint programming
Dynamic scoring functions with variable expressions: new SLS methods for solving SAT
SAT'10 Proceedings of the 13th international conference on Theory and Applications of Satisfiability Testing
Automated configuration of mixed integer programming solvers
CPAIOR'10 Proceedings of the 7th international conference on Integration of AI and OR Techniques in Constraint Programming for Combinatorial Optimization Problems
Sequential model-based optimization for general algorithm configuration
LION'05 Proceedings of the 5th international conference on Learning and Intelligent Optimization
Automatically configuring algorithms for scaling performance
LION'12 Proceedings of the 6th international conference on Learning and Intelligent Optimization
Hi-index | 0.00 |
Automated algorithm configuration has been proven to be an effective approach for achieving improved performance of solvers for many computationally hard problems. We consider the challenging situation where the kind of problem instances for which we desire optimised performance is too difficult to be used during the configuration process. Here, we propose a novel combination of racing techniques with existing algorithm configurators to meet this challenge. We demonstrate that, applied to state-of-the-art solver for propositional satisfiability, mixed integer programming and travelling salesman problems, the resulting algorithm configuration protocol achieves better results than previous approaches and in many cases closely matches the bound on performance obtained using an oracle selector. We also report results indicating that the performance of our new racing protocols is quite robust to variations in the confidence level of the test used for eliminating weak configurations, and that performance benefits from presenting instances ordered according to increasing difficulty during the race -- something not done in standard racing procedures.