ReACTR: Realtime Algorithm Configuration through Tournament Rankings
Refereed Conference Meeting Proceeding
It is now readily accepted that automated algorithm configuration is a necessity for ensuring optimized performance of solvers on a particular domain. Even the best researchers, who have designed their solver from the ground up, are not always able to manually find the best parameter settings. This has been repeatedly demonstrated by configuration tools like ParamILS, SMAC, and GGA. However, all these techniques currently assume a static environment, where demonstrative instances are provided beforehand, potentially unlimited time is provided to adequately search the parameter space, and the solver would never need to be retrained. This is not always the case in practice. The ReACT paper, demonstrated that an solver could be configured during runtime as new instances arrive at a steady stream. This paper expands on this approach and shows how a ranking approach like TrueSkill can further improve the configurator's performance, making it able to quickly find good parameterizations without adding any overhead on the time needed to solve any new instance, and then continuously improve as new instances are evaluated. The modifications to ReACT we present even let us outperform existing static configurators like SMAC in a non-dynamic setting.
International Joint Conference on Artificial Intelligence
Proceedings of the Twenty-Fourth International Joint Conference on Artificial Intelligence
Digital Object Identifer (DOI):
National University of Ireland, Cork (UCC)
Open access repository: