@Proceedings{OASC2017,
title = {Proceedings of the Open Algorithm Selection Challenge},
booktitle = {Proceedings of the Open Algorithm Selection Challenge},
editor = {Marius Lindauer and Jan N. van Rijn and Lars Kotthoff},
publisher = {PMLR},
series = {Proceedings of Machine Learning Research},
volume = 79
}
@InProceedings{pmlr-v79-lindauer17a,
title = {Open Algorithm Selection Challenge 2017: Setup and Scenarios},
author = {Lindauer, Marius and van Rijn, Jan N. and Kotthoff, Lars},
booktitle = {Proceedings of the Open Algorithm Selection Challenge},
pages = {1--7},
year = {2017},
editor = {Lindauer, Marius and van Rijn, Jan N. and Kotthoff, Lars},
volume = {79},
series = {Proceedings of Machine Learning Research},
month = {11--12 Sep},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v79/lindauer17a/lindauer17a.pdf},
url = {https://proceedings.mlr.press/v79/lindauer17a.html},
abstract = {The 2017 algorithm selection challenge provided a snapshot of the state of the art in algorithm selection and garnered submissions from four teams. In this chapter, we describe the setup of the challenge and the algorithm scenarios that were used.}
}
@InProceedings{pmlr-v79-gonard17a,
title = {ASAP.V2 and ASAP.V3: Sequential optimization of an Algorithm Selector and a Scheduler},
author = {Gonard, François and Schoenauer, Marc and Sebag, Michèle},
booktitle = {Proceedings of the Open Algorithm Selection Challenge},
pages = {8--11},
year = {2017},
editor = {Lindauer, Marius and van Rijn, Jan N. and Kotthoff, Lars},
volume = {79},
series = {Proceedings of Machine Learning Research},
month = {11--12 Sep},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v79/gonard17a/gonard17a.pdf},
url = {https://proceedings.mlr.press/v79/gonard17a.html},
abstract = {Algorithm portfolios are known to offer robust performances, efficiently overcoming the weakness of every single algorithm on some particular problem instances. The presented asap system relies on the alternate optimization of two complementary portfolio approaches, namely a sequential scheduler and a per-instance algorithm selector.}
}
@InProceedings{pmlr-v79-liu17a,
title = {SUNNY with Algorithm Configuration},
author = {Liu, Tong and Amadini, Roberto and Mauro, Jacopo},
booktitle = {Proceedings of the Open Algorithm Selection Challenge},
pages = {12--14},
year = {2017},
editor = {Lindauer, Marius and van Rijn, Jan N. and Kotthoff, Lars},
volume = {79},
series = {Proceedings of Machine Learning Research},
month = {11--12 Sep},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v79/liu17a/liu17a.pdf},
url = {https://proceedings.mlr.press/v79/liu17a.html},
abstract = {The SUNNY algorithm is a portfolio technique originally tailored for Constraint Satisfaction Problems (CSPs). SUNNY allows to select a set of solvers to be run on a given CSP, and was proven to be effective in the MiniZinc Challenge, i.e., the yearly international competition for CP solvers. In 2015, SUNNY was compared with other solver selectors in the first ICON Challenge on algorithm selection with less satisfactory performance. In this paper we briefly describe the new version of the SUNNY approach for algorithm selection, that was submitted to the first Open Algorithm Selection Challenge.}
}
@InProceedings{pmlr-v79-cameron17a,
title = {OASC-2017: *Zilla Submission},
author = {Cameron, Chris and Hoos, Holger H. and Leyton-Brown, Kevin and Hutter, Frank},
booktitle = {Proceedings of the Open Algorithm Selection Challenge},
pages = {15--18},
year = {2017},
editor = {Lindauer, Marius and van Rijn, Jan N. and Kotthoff, Lars},
volume = {79},
series = {Proceedings of Machine Learning Research},
month = {11--12 Sep},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v79/cameron17a/cameron17a.pdf},
url = {https://proceedings.mlr.press/v79/cameron17a.html},
abstract = {*Zilla is a model-based approach for algorithm selection and the most recent iteration of the well-known SATzilla project. The new *Zilla system has increased flexibility for the user and is configurable to run with many machine learning models and alternatives for various pre/post processing steps (e.g., presolver selection, feature completion prediction, and solver subset selection). The main additions to our *Zilla pipeline are automated procedures for feature group selection, hyper-parameter tuning, and solver subsampling prior to model building. We submit two versions for the competition that are equivalent except for the choice of per-instance machine learning model. For our first submission, we use a weighted pairwise random forest classifier. For our second submission, we test an experimental approach that offline, builds a weighted pairwise random forest classifier and online, finds the nearest instances based on the average path lengths across trees and optimizes a schedule over those instances.}
}
@InProceedings{pmlr-v79-malone17a,
title = {AS-ASL: Algorithm Selection with Auto-sklearn},
author = {Malone, Brandon and Kangas, Kustaa and Järvisalo, Matti and Koivisto, Mikko and Myllymäki, Petri},
booktitle = {Proceedings of the Open Algorithm Selection Challenge},
pages = {19--22},
year = {2017},
editor = {Lindauer, Marius and van Rijn, Jan N. and Kotthoff, Lars},
volume = {79},
series = {Proceedings of Machine Learning Research},
month = {11--12 Sep},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v79/malone17a/malone17a.pdf},
url = {https://proceedings.mlr.press/v79/malone17a.html},
abstract = {In this paper, we describe our algorithm selection with Auto-sklearn (as-asl) software as it was entered in the 2017 Open Algorithm Selection Challenge. as-asl first selects informative sets of features and then uses those to predict distributions of algorithm runtimes. A classifier uses those predictions, as well as the informative features, to select an algorithm for each problem instance. Our source code is publicly available with the permissive MIT license.}
}