[{"date_updated":"2022-01-06T06:54:42Z","publication_status":"published","title":"AutoML for Multi-Label Classification: Overview and Empirical Evaluation","year":"2021","status":"public","author":[{"first_name":"Marcel Dominik","last_name":"Wever","orcid":" https://orcid.org/0000-0001-9782-6818","full_name":"Wever, Marcel Dominik","id":"33176"},{"full_name":"Tornede, Alexander","first_name":"Alexander","last_name":"Tornede","id":"38209"},{"full_name":"Mohr, Felix","last_name":"Mohr","first_name":"Felix"},{"id":"48129","full_name":"Hüllermeier, Eyke","last_name":"Hüllermeier","first_name":"Eyke"}],"publication_identifier":{"issn":["0162-8828","2160-9292","1939-3539"]},"doi":"10.1109/tpami.2021.3051276","user_id":"5786","page":"1-1","_id":"21004","language":[{"iso":"eng"}],"abstract":[{"lang":"eng","text":"Automated machine learning (AutoML) supports the algorithmic construction and data-specific customization of machine learning pipelines, including the selection, combination, and parametrization of machine learning algorithms as main constituents. Generally speaking, AutoML approaches comprise two major components: a search space model and an optimizer for traversing the space. Recent approaches have shown impressive results in the realm of supervised learning, most notably (single-label) classification (SLC). Moreover, first attempts at extending these approaches towards multi-label classification (MLC) have been made. While the space of candidate pipelines is already huge in SLC, the complexity of the search space is raised to an even higher power in MLC. One may wonder, therefore, whether and to what extent optimizers established for SLC can scale to this increased complexity, and how they compare to each other. This paper makes the following contributions: First, we survey existing approaches to AutoML for MLC. Second, we augment these approaches with optimizers not previously tried for MLC. Third, we propose a benchmarking framework that supports a fair and systematic comparison. Fourth, we conduct an extensive experimental study, evaluating the methods on a suite of MLC problems. We find a grammar-based best-first search to compare favorably to other optimizers."}],"project":[{"_id":"1","name":"SFB 901"},{"name":"SFB 901 - Project Area B","_id":"3"},{"name":"SFB 901 - Subproject B2","_id":"10"},{"_id":"52","name":"Computing Resources Provided by the Paderborn Center for Parallel Computing"}],"publication":"IEEE Transactions on Pattern Analysis and Machine Intelligence","citation":{"chicago":"Wever, Marcel Dominik, Alexander Tornede, Felix Mohr, and Eyke Hüllermeier. “AutoML for Multi-Label Classification: Overview and Empirical Evaluation.” <i>IEEE Transactions on Pattern Analysis and Machine Intelligence</i>, 2021, 1–1. <a href=\"https://doi.org/10.1109/tpami.2021.3051276\">https://doi.org/10.1109/tpami.2021.3051276</a>.","short":"M.D. Wever, A. Tornede, F. Mohr, E. Hüllermeier, IEEE Transactions on Pattern Analysis and Machine Intelligence (2021) 1–1.","ama":"Wever MD, Tornede A, Mohr F, Hüllermeier E. AutoML for Multi-Label Classification: Overview and Empirical Evaluation. <i>IEEE Transactions on Pattern Analysis and Machine Intelligence</i>. Published online 2021:1-1. doi:<a href=\"https://doi.org/10.1109/tpami.2021.3051276\">10.1109/tpami.2021.3051276</a>","bibtex":"@article{Wever_Tornede_Mohr_Hüllermeier_2021, title={AutoML for Multi-Label Classification: Overview and Empirical Evaluation}, DOI={<a href=\"https://doi.org/10.1109/tpami.2021.3051276\">10.1109/tpami.2021.3051276</a>}, journal={IEEE Transactions on Pattern Analysis and Machine Intelligence}, author={Wever, Marcel Dominik and Tornede, Alexander and Mohr, Felix and Hüllermeier, Eyke}, year={2021}, pages={1–1} }","apa":"Wever, M. D., Tornede, A., Mohr, F., &#38; Hüllermeier, E. (2021). AutoML for Multi-Label Classification: Overview and Empirical Evaluation. <i>IEEE Transactions on Pattern Analysis and Machine Intelligence</i>, 1–1. <a href=\"https://doi.org/10.1109/tpami.2021.3051276\">https://doi.org/10.1109/tpami.2021.3051276</a>","mla":"Wever, Marcel Dominik, et al. “AutoML for Multi-Label Classification: Overview and Empirical Evaluation.” <i>IEEE Transactions on Pattern Analysis and Machine Intelligence</i>, 2021, pp. 1–1, doi:<a href=\"https://doi.org/10.1109/tpami.2021.3051276\">10.1109/tpami.2021.3051276</a>.","ieee":"M. D. Wever, A. Tornede, F. Mohr, and E. Hüllermeier, “AutoML for Multi-Label Classification: Overview and Empirical Evaluation,” <i>IEEE Transactions on Pattern Analysis and Machine Intelligence</i>, pp. 1–1, 2021, doi: <a href=\"https://doi.org/10.1109/tpami.2021.3051276\">10.1109/tpami.2021.3051276</a>."},"type":"journal_article","keyword":["Automated Machine Learning","Multi Label Classification","Hierarchical Planning","Bayesian Optimization"],"department":[{"_id":"34"},{"_id":"355"},{"_id":"26"}],"date_created":"2021-01-16T14:48:13Z"},{"file":[{"creator":"ups","date_created":"2018-11-02T15:32:16Z","relation":"main_file","date_updated":"2018-11-02T15:32:16Z","file_name":"ML-PlanAutomatedMachineLearnin.pdf","file_size":1070937,"access_level":"closed","file_id":"5306","content_type":"application/pdf","success":1}],"date_created":"2018-07-08T14:06:14Z","type":"journal_article","keyword":["AutoML","Hierarchical Planning","HTN planning","ML-Plan"],"department":[{"_id":"355"},{"_id":"34"},{"_id":"7"},{"_id":"26"}],"publication":"Machine Learning","abstract":[{"lang":"eng","text":"Automated machine learning (AutoML) seeks to automatically select, compose, and parametrize machine learning algorithms, so as to achieve optimal performance on a given task (dataset). Although current approaches to AutoML have already produced impressive results, the field is still far from mature, and new techniques are still being developed. In this paper, we present ML-Plan, a new approach to AutoML based on hierarchical planning. To highlight the potential of this approach, we compare ML-Plan to the state-of-the-art frameworks Auto-WEKA, auto-sklearn, and TPOT. In an extensive series of experiments, we show that ML-Plan is highly competitive and often outperforms existing approaches."}],"main_file_link":[{"open_access":"1","url":"https://rdcu.be/3Nc2"}],"language":[{"iso":"eng"}],"doi":"10.1007/s10994-018-5735-z","title":"ML-Plan: Automated Machine Learning via Hierarchical Planning","year":"2018","author":[{"full_name":"Mohr, Felix","first_name":"Felix","last_name":"Mohr"},{"id":"33176","full_name":"Wever, Marcel Dominik","first_name":"Marcel Dominik","orcid":" https://orcid.org/0000-0001-9782-6818","last_name":"Wever"},{"id":"48129","full_name":"Hüllermeier, Eyke","last_name":"Hüllermeier","first_name":"Eyke"}],"publication_identifier":{"eissn":["1573-0565"],"issn":["0885-6125"]},"date_updated":"2022-01-06T06:59:21Z","publication_status":"epub_ahead","article_type":"original","oa":"1","file_date_updated":"2018-11-02T15:32:16Z","citation":{"mla":"Mohr, Felix, et al. “ML-Plan: Automated Machine Learning via Hierarchical Planning.” <i>Machine Learning</i>, Springer, 2018, pp. 1495–515, doi:<a href=\"https://doi.org/10.1007/s10994-018-5735-z\">10.1007/s10994-018-5735-z</a>.","apa":"Mohr, F., Wever, M. D., &#38; Hüllermeier, E. (2018). ML-Plan: Automated Machine Learning via Hierarchical Planning. <i>Machine Learning</i>, 1495–1515. <a href=\"https://doi.org/10.1007/s10994-018-5735-z\">https://doi.org/10.1007/s10994-018-5735-z</a>","ieee":"F. Mohr, M. D. Wever, and E. Hüllermeier, “ML-Plan: Automated Machine Learning via Hierarchical Planning,” <i>Machine Learning</i>, pp. 1495–1515, 2018, doi: <a href=\"https://doi.org/10.1007/s10994-018-5735-z\">10.1007/s10994-018-5735-z</a>.","chicago":"Mohr, Felix, Marcel Dominik Wever, and Eyke Hüllermeier. “ML-Plan: Automated Machine Learning via Hierarchical Planning.” <i>Machine Learning</i>, 2018, 1495–1515. <a href=\"https://doi.org/10.1007/s10994-018-5735-z\">https://doi.org/10.1007/s10994-018-5735-z</a>.","short":"F. Mohr, M.D. Wever, E. Hüllermeier, Machine Learning (2018) 1495–1515.","ama":"Mohr F, Wever MD, Hüllermeier E. ML-Plan: Automated Machine Learning via Hierarchical Planning. <i>Machine Learning</i>. Published online 2018:1495-1515. doi:<a href=\"https://doi.org/10.1007/s10994-018-5735-z\">10.1007/s10994-018-5735-z</a>","bibtex":"@article{Mohr_Wever_Hüllermeier_2018, title={ML-Plan: Automated Machine Learning via Hierarchical Planning}, DOI={<a href=\"https://doi.org/10.1007/s10994-018-5735-z\">10.1007/s10994-018-5735-z</a>}, journal={Machine Learning}, publisher={Springer}, author={Mohr, Felix and Wever, Marcel Dominik and Hüllermeier, Eyke}, year={2018}, pages={1495–1515} }"},"project":[{"_id":"1","name":"SFB 901"},{"_id":"3","name":"SFB 901 - Project Area B"},{"_id":"10","name":"SFB 901 - Subproject B2"},{"name":"Computing Resources Provided by the Paderborn Center for Parallel Computing","_id":"52"}],"page":"1495-1515","_id":"3510","publisher":"Springer","ddc":["000"],"user_id":"5786","status":"public","conference":{"location":"Dublin, Ireland","start_date":"2018-09-10","name":"European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases","end_date":"2018-09-14"},"has_accepted_license":"1"},{"publication":"ICML 2018 AutoML Workshop","abstract":[{"text":"In automated machine learning (AutoML), the process of engineering machine learning applications with respect to a specific problem is (partially) automated.\r\nVarious AutoML tools have already been introduced to provide out-of-the-box machine learning functionality.\r\nMore specifically, by selecting machine learning algorithms and optimizing their hyperparameters, these tools produce a machine learning pipeline tailored to the problem at hand.\r\nExcept for TPOT, all of these tools restrict the maximum number of processing steps of such a pipeline.\r\nHowever, as TPOT follows an evolutionary approach, it suffers from performance issues when dealing with larger datasets.\r\nIn this paper, we present an alternative approach leveraging a hierarchical planning to configure machine learning pipelines that are unlimited in length.\r\nWe evaluate our approach and find its performance to be competitive with other AutoML tools, including TPOT.","lang":"eng"}],"date_created":"2018-08-09T06:14:54Z","file":[{"file_name":"38.pdf","access_level":"open_access","file_size":297811,"relation":"main_file","date_updated":"2018-08-09T06:14:43Z","file_id":"3853","content_type":"application/pdf","creator":"wever","date_created":"2018-08-09T06:14:43Z"}],"department":[{"_id":"355"}],"type":"conference","keyword":["automated machine learning","complex pipelines","hierarchical planning"],"author":[{"id":"33176","full_name":"Wever, Marcel Dominik","orcid":" https://orcid.org/0000-0001-9782-6818","first_name":"Marcel Dominik","last_name":"Wever"},{"full_name":"Mohr, Felix","last_name":"Mohr","first_name":"Felix"},{"id":"48129","first_name":"Eyke","last_name":"Hüllermeier","full_name":"Hüllermeier, Eyke"}],"title":"ML-Plan for Unlimited-Length Machine Learning Pipelines","year":"2018","date_updated":"2022-01-06T06:59:46Z","language":[{"iso":"eng"}],"main_file_link":[{"url":"https://docs.google.com/viewer?a=v&pid=sites&srcid=ZGVmYXVsdGRvbWFpbnxhdXRvbWwyMDE4aWNtbHxneDo3M2Q3MjUzYjViNDRhZTAx"}],"citation":{"ama":"Wever MD, Mohr F, Hüllermeier E. ML-Plan for Unlimited-Length Machine Learning Pipelines. In: <i>ICML 2018 AutoML Workshop</i>. ; 2018.","bibtex":"@inproceedings{Wever_Mohr_Hüllermeier_2018, title={ML-Plan for Unlimited-Length Machine Learning Pipelines}, booktitle={ICML 2018 AutoML Workshop}, author={Wever, Marcel Dominik and Mohr, Felix and Hüllermeier, Eyke}, year={2018} }","mla":"Wever, Marcel Dominik, et al. “ML-Plan for Unlimited-Length Machine Learning Pipelines.” <i>ICML 2018 AutoML Workshop</i>, 2018.","short":"M.D. Wever, F. Mohr, E. Hüllermeier, in: ICML 2018 AutoML Workshop, 2018.","chicago":"Wever, Marcel Dominik, Felix Mohr, and Eyke Hüllermeier. “ML-Plan for Unlimited-Length Machine Learning Pipelines.” In <i>ICML 2018 AutoML Workshop</i>, 2018.","apa":"Wever, M. D., Mohr, F., &#38; Hüllermeier, E. (2018). ML-Plan for Unlimited-Length Machine Learning Pipelines. In <i>ICML 2018 AutoML Workshop</i>. Stockholm, Sweden.","ieee":"M. D. Wever, F. Mohr, and E. Hüllermeier, “ML-Plan for Unlimited-Length Machine Learning Pipelines,” in <i>ICML 2018 AutoML Workshop</i>, Stockholm, Sweden, 2018."},"file_date_updated":"2018-08-09T06:14:43Z","project":[{"_id":"1","name":"SFB 901"},{"name":"SFB 901 - Project Area B","_id":"3"},{"_id":"10","name":"SFB 901 - Subproject B2"}],"quality_controlled":"1","oa":"1","conference":{"location":"Stockholm, Sweden","name":"ICML 2018 AutoML Workshop","start_date":"2018-07-10","end_date":"2018-07-15"},"status":"public","has_accepted_license":"1","urn":"38527","_id":"3852","user_id":"49109","ddc":["006"]}]