{"status":"public","date_created":"2022-11-17T12:57:40Z","external_id":{"arxiv":["2210.17341"]},"citation":{"short":"L. Fehring, J.M. Hanselle, A. Tornede, in: Workshop on Meta-Learning (MetaLearn 2022) @ NeurIPS 2022, 2022.","bibtex":"@inproceedings{Fehring_Hanselle_Tornede_2022, title={HARRIS: Hybrid Ranking and Regression Forests for Algorithm Selection}, booktitle={Workshop on Meta-Learning (MetaLearn 2022) @ NeurIPS 2022}, author={Fehring, Lukass and Hanselle, Jonas Manuel and Tornede, Alexander}, year={2022} }","chicago":"Fehring, Lukass, Jonas Manuel Hanselle, and Alexander Tornede. “HARRIS: Hybrid Ranking and Regression Forests for Algorithm Selection.” In *Workshop on Meta-Learning (MetaLearn 2022) @ NeurIPS 2022*, 2022.","mla":"Fehring, Lukass, et al. “HARRIS: Hybrid Ranking and Regression Forests for Algorithm Selection.” *Workshop on Meta-Learning (MetaLearn 2022) @ NeurIPS 2022*, 2022.","ieee":"L. Fehring, J. M. Hanselle, and A. Tornede, “HARRIS: Hybrid Ranking and Regression Forests for Algorithm Selection,” presented at the Workshop on Meta-Learning (MetaLearn 2022) @ NeurIPS 2022, Baltimore, 2022.","ama":"Fehring L, Hanselle JM, Tornede A. HARRIS: Hybrid Ranking and Regression Forests for Algorithm Selection. In: *Workshop on Meta-Learning (MetaLearn 2022) @ NeurIPS 2022*. ; 2022.","apa":"Fehring, L., Hanselle, J. M., & Tornede, A. (2022). HARRIS: Hybrid Ranking and Regression Forests for Algorithm Selection. *Workshop on Meta-Learning (MetaLearn 2022) @ NeurIPS 2022*. Workshop on Meta-Learning (MetaLearn 2022) @ NeurIPS 2022, Baltimore."},"user_id":"38209","publication":"Workshop on Meta-Learning (MetaLearn 2022) @ NeurIPS 2022","author":[{"full_name":"Fehring, Lukass","first_name":"Lukass","last_name":"Fehring"},{"first_name":"Jonas Manuel","full_name":"Hanselle, Jonas Manuel","orcid":"0000-0002-1231-4985","id":"43980","last_name":"Hanselle"},{"last_name":"Tornede","id":"38209","first_name":"Alexander","full_name":"Tornede, Alexander"}],"type":"conference","_id":"34103","language":[{"iso":"eng"}],"year":"2022","conference":{"name":"Workshop on Meta-Learning (MetaLearn 2022) @ NeurIPS 2022","location":"Baltimore"},"abstract":[{"text":"It is well known that different algorithms perform differently well on an\r\ninstance of an algorithmic problem, motivating algorithm selection (AS): Given\r\nan instance of an algorithmic problem, which is the most suitable algorithm to\r\nsolve it? As such, the AS problem has received considerable attention resulting\r\nin various approaches - many of which either solve a regression or ranking\r\nproblem under the hood. Although both of these formulations yield very natural\r\nways to tackle AS, they have considerable weaknesses. On the one hand,\r\ncorrectly predicting the performance of an algorithm on an instance is a\r\nsufficient, but not a necessary condition to produce a correct ranking over\r\nalgorithms and in particular ranking the best algorithm first. On the other\r\nhand, classical ranking approaches often do not account for concrete\r\nperformance values available in the training data, but only leverage rankings\r\ncomposed from such data. We propose HARRIS- Hybrid rAnking and RegRessIon\r\nforeSts - a new algorithm selector leveraging special forests, combining the\r\nstrengths of both approaches while alleviating their weaknesses. HARRIS'\r\ndecisions are based on a forest model, whose trees are created based on splits\r\noptimized on a hybrid ranking and regression loss function. As our preliminary\r\nexperimental study on ASLib shows, HARRIS improves over standard algorithm\r\nselection approaches on some scenarios showing that combining ranking and\r\nregression in trees is indeed promising for AS.","lang":"eng"}],"date_updated":"2022-11-17T13:00:53Z","title":"HARRIS: Hybrid Ranking and Regression Forests for Algorithm Selection","project":[{"name":"SFB 901: SFB 901","_id":"1"},{"name":"SFB 901 - B: SFB 901 - Project Area B","_id":"3"},{"name":"SFB 901 - B2: SFB 901 - Subproject B2","_id":"10"}]}