{"type":"ScholarlyArticle","title":"Automated Machine Learning with Monte-Carlo Tree Search","authors":[{"name":"Herilalaina Rakotoarison","slug":"herilalaina-rakotoarison","given":"Herilalaina","family":"Rakotoarison","affiliation":"TAU , LRI-CNRS - INRIA Université Paris-Saclay , France"},{"name":"Marc Schoenauer","slug":"marc-schoenauer","given":"Marc","family":"Schoenauer","affiliation":"TAU , LRI-CNRS - INRIA Université Paris-Saclay , France"},{"name":"Michèle Sebag","slug":"michele-sebag","given":"Michèle","family":"Sebag","affiliation":"TAU , LRI-CNRS - INRIA Université Paris-Saclay , France"}],"abstract":"The AutoML approach aims to deliver peak performance from a machine learning  portfolio on the dataset at hand. A Monte-Carlo Tree Search Algorithm Selection and Configuration (Mosaic) approach is presented to tackle this mixed (combinatorial and continuous) expensive optimization problem on the structured search space of ML pipelines. Extensive lesion studies are conducted to independently assess and compare: i) the optimization processes based on Bayesian Optimization or Monte Carlo Tree Search (MCTS); ii) its warm-start initialization based on meta-features or random runs; iii) the ensembling of the solutions gathered along the search. Mosaic is assessed on the OpenML 100 benchmark and the Scikit-learn portfolio, with statistically significant gains over AutoSkLearn, winner of all former AutoML challenges.","doi":"10.24963/ijcai.2019/457","url":"https://ijcai.org/proceedings/2019/457","pdf":"https://ijcai.org/proceedings/2019/0457.pdf","markdown":"https://ijcai.org/proceedings/2019/457.md","bibtex":"https://ijcai.org/proceedings/2019/bibtex/457","volume":{"slug":"2019","title":"Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2019","url":"https://ijcai.org/proceedings/2019/","year":2019,"location":"Macao","dates":"10-16 August 2019","editors":"Sarit Kraus","isbn":"978-0-9992411-4-1","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2019"},"section":"Machine Learning M-Z","track":"Main track","number":457,"pages":{"first":3296,"last":3303},"keywords":["Uncertainty in AI: Sequential Decision Making","Machine Learning: Ensemble Methods","Machine Learning: Classification"],"videos":[],"references":[{"position":1,"raw":"[Auer, 2002] P. Auer. Using Confidence Bounds for Exploitation-Exploration Trade-offs. Journal of Machine Learning Research, 3:397-422, 2002.","title":"Using Confidence Bounds for Exploitation-Exploration Trade-offs","authors":"P Auer","year":"2002","venue":"Journal of Machine Learning Research","doi":null,"url":null,"internal_url":null},{"position":2,"raw":"[Auger et al., 2013] David Auger, Adrien Couetoux, and Olivier Teytaud. Continuous upper confidence trees with polynomial exploration-consistency. In ECML-KDD, pages 194-209. Springer, 2013.","title":"Continuous upper confidence trees with polynomial exploration-consistency","authors":"David Auger, Adrien Couetoux, Olivier Teytaud","year":"2013","venue":"ECML-KDD","doi":null,"url":null,"internal_url":null},{"position":3,"raw":"[Balaji and Allen, 2018] A. Balaji and A. Allen. Bench- marking Automatic Machine Learning Frameworks. arXiv:1808.06492 [cs, stat], 2018.","title":"Benchmarking Automatic Machine Learning Frameworks","authors":"A Balaji, A Allen","year":"2018","venue":null,"doi":null,"url":null,"internal_url":null},{"position":4,"raw":"[Bergstra et al., 2011] J. S. Bergstra, R. Bardenet, Y. Bengio, and B. Kégl. Algorithms for Hyper-Parameter Optimiza- tion. In NIPS, pages 2546-2554, 2011.","title":"Algorithms for Hyper-Parameter Optimization","authors":"J S Bergstra, R Bardenet, Y Bengio, B Kégl","year":"2011","venue":"NIPS","doi":null,"url":null,"internal_url":null},{"position":5,"raw":"[Brazdil and Giraud-Carrier, 2018] P. Brazdil and Ch. Giraud-Carrier. Metalearning and Algorithm Selection: Progress, State of the Art and Introduction to the Special Issue. Machine Learning, 107(1):1-14, 2018.","title":"Metalearning and Algorithm Selection: Progress, State of the Art and Introduction to the Special Issue","authors":"P Brazdil, Ch Giraud-Carrier","year":"2018","venue":"Machine Learning","doi":null,"url":null,"internal_url":null},{"position":6,"raw":"[Bubeck et al., 2011] S. Bubeck, R. Munos, G. Stoltz, and C. Szepesvári. X-armed bandits. Journal of Machine Learning Research, 12:1655-1695, 2011.","title":"X-armed bandits","authors":"S Bubeck, R Munos, G Stoltz, C Szepesvári","year":"2011","venue":"Journal of Machine Learning Research","doi":null,"url":null,"internal_url":null},{"position":7,"raw":"[Caruana et al., 2004] R. Caruana, A. Niculescu-Mizil, G. Crew, and A. Ksikes. Ensemble selection from libraries of models. In Proc.ICML, page 18, 2004.","title":"Ensemble selection from libraries of models","authors":"R Caruana, A Niculescu-Mizil, G Crew, A Ksikes","year":"2004","venue":"Proc.ICML","doi":null,"url":null,"internal_url":null},{"position":8,"raw":"[Chaslot et al., 2008] G. Chaslot, S. Bakkes, I. Szita, and P. Spronck. Monte-Carlo Tree Search: A New Framework for Game AI. In AI and Interactive Digital Entertainment, pages 216-217. AAAI Press, 2008.","title":"Monte-Carlo Tree Search: A New Framework for Game AI","authors":"G Chaslot, S Bakkes, I Szita, P Spronck","year":"2008","venue":"AI and Interactive Digital Entertainment","doi":null,"url":null,"internal_url":null},{"position":9,"raw":"[Chen et al., 2018] B. Chen, H. Wu, W. Mo, I. Chattopad- hyay, and H. Lipson. Autostacker: A Compositional Evo- lutionary Learning System. In Proc. ACM-GECCO, pages 402-409. ACM Press, 2018.","title":"Autostacker: A Compositional Evolutionary Learning System","authors":"B Chen, H Wu, W Mo, I Chattopadhyay, H Lipson","year":"2018","venue":"Proc. ACM-GECCO","doi":null,"url":null,"internal_url":null},{"position":10,"raw":"[Cohen and Welling, 2016] T. Cohen and M. Welling. Group Equivariant Convolutional Networks. In Proc. ICML, vol- ume 48, pages 2990-2999. PMLR, 2016.","title":"Group Equivariant Convolutional Networks","authors":"T Cohen, M Welling","year":"2016","venue":"Proc. ICML","doi":null,"url":null,"internal_url":null},{"position":11,"raw":"[Coquelin and Munos, 2007] P.-A. Coquelin and R. Munos. Bandit algorithms for tree search. In Proc. UAI 2007, pages 67-74, 2007.","title":"Bandit algorithms for tree search","authors":"P.-A Coquelin, R Munos","year":"2007","venue":"Proc. UAI 2007","doi":null,"url":null,"internal_url":null},{"position":12,"raw":"[Drori et al., 2018] I. Drori, Y. Krishnamurthy, et al. Al- phaD3M: Machine Learning Pipeline Synthesis . In ICML workshop on AutoML, 2018.","title":"Al-phaD3M: Machine Learning Pipeline Synthesis","authors":"Drori","year":"2018","venue":"ICML workshop on AutoML","doi":null,"url":null,"internal_url":null},{"position":13,"raw":"[Eggensperger et al., 2013] K. Eggensperger, M. Feurer, F. Hutter, et al. Towards an Empirical Foundation for Assessing Bayesian Optimization of Hyperparameters. In NIPS wkp on BO in Theory and Practice, 2013.","title":"Towards an Empirical Foundation for Assessing Bayesian Optimization of Hyperparameters","authors":"K Eggensperger, M Feurer, F Hutter","year":"2013","venue":"NIPS wkp on BO in Theory and Practice","doi":null,"url":null,"internal_url":null},{"position":14,"raw":"[Feurer et al., 2015] M. Feurer, Klein A., Eggensperger K., M. Blum, and F. Hutter. Efficient and Robust Automated Machine Learning. In C. Cortes et al., editor, NIPS 28, pages 2962-2970, 2015.","title":"Efficient and Robust Automated Machine Learning","authors":"M Feurer, A Klein, K Eggensperger, M Blum, F Hutter","year":"2015","venue":"NIPS","doi":null,"url":null,"internal_url":null},{"position":15,"raw":"[Gelly and Silver, 2011] S. Gelly and D. Silver. Monte-Carlo Tree Search and Rapid Action Value Estimation in Com- puter GO. Artificial Intelligence, 175:1856-1875, 2011.","title":"Monte-Carlo Tree Search and Rapid Action Value Estimation in Computer GO","authors":"S Gelly, D Silver","year":"2011","venue":"Artificial Intelligence","doi":null,"url":null,"internal_url":null},{"position":16,"raw":"[Guyon et al., 2015] I. Guyon, K. Bennett, G. Cawley, et al. Design of the 2015 Chalearn AutoML challenge. In Proc. IJCNN, pages 1-8. IEEE, 2015.","title":"Design of the 2015 Chalearn AutoML challenge","authors":"Guyon","year":"2015","venue":"Proc. IJCNN","doi":null,"url":null,"internal_url":null},{"position":17,"raw":"[Guyon et al., 2018] I. Guyon, W.-W. Tu, et al. Automatic Machine Learning Challenge 2018: Towards AI for Ev- eryone. In PAKDD 2018 Data Competition, 2018.","title":"Automatic Machine Learning Challenge 2018: Towards AI for Everyone","authors":"Guyon","year":"2018","venue":"PAKDD 2018 Data Competition","doi":null,"url":null,"internal_url":null},{"position":18,"raw":"[Hutter et al., 2011] F. Hutter, H.H. Hoos, and K. Leyton- Brown. Sequential Model-based Optimization for General Algorithm Configuration. In Proc. LION, pages 507-523. Springer Verlag, 2011.","title":"Sequential Model-based Optimization for General Algorithm Configuration","authors":"F Hutter, H H Hoos, K Leyton-Brown","year":"2011","venue":"Proc. LION","doi":null,"url":null,"internal_url":null},{"position":19,"raw":"[Klein et al., 2017] A. Klein, S. Falkner, et al. Fast Bayesian Optimization of Machine Learning Hyperparameters on Large Datasets. In Proc. AISTAT, pages 528-536, 2017.","title":"Fast Bayesian Optimization of Machine Learning Hyperparameters on Large Datasets","authors":"A Klein, S Falkner","year":"2017","venue":"Proc. AISTAT","doi":null,"url":null,"internal_url":null},{"position":20,"raw":"[Kocsis and Szepesvári, 2006] L. Kocsis and C. Szepesvári. Bandit Based Monte-Carlo Planning. In Fürnkranz et al., editor, Proc. ECML, pages 282-293. Springer, 2006.","title":"Bandit Based Monte-Carlo Planning","authors":"L Kocsis, C Szepesvári","year":"2006","venue":"Proc. ECML","doi":null,"url":null,"internal_url":null},{"position":21,"raw":"[Kotthoff et al., 2017] L. Kotthoff, Ch. Thornton, H.H. Hoos, F. Hutter, and K. Leyton-Brown. Auto-WEKA 2.0: Automatic Model Selection and Hyperparameter Op- timization in WEKA. JMLR, 18(25):1-5, 2017.","title":"Auto-WEKA 2.0: Automatic Model Selection and Hyperparameter Optimization in WEKA","authors":"L Kotthoff, Ch Thornton, H H Hoos, F Hutter, K Leyton-Brown","year":"2017","venue":"JMLR","doi":null,"url":null,"internal_url":null},{"position":22,"raw":"[Li et al., 2017] L. Li, K. Jamieson, G. DeSalvo, et al. Hy- perband: A novel bandit-based approach to hyperparame- ter optimization. JMLR, 18(1):6765-6816, 2017.","title":"Hyperband: A novel bandit-based approach to hyperparameter optimization","authors":"L Li, K Jamieson, G Desalvo","year":"2017","venue":"JMLR","doi":null,"url":null,"internal_url":null},{"position":23,"raw":"[Mockus et al., 1978] J. Mockus, V. Tiesis, and A. Zilinskas. The application of Bayesian methods for seeking the ex- tremum. In L. Dixon and G. Szego, editors, Toward Global Optimization, volume 2. Elsevier, 1978.","title":"The application of Bayesian methods for seeking the extremum","authors":"J Mockus, V Tiesis, A Zilinskas","year":"1978","venue":"Toward Global Optimization","doi":null,"url":null,"internal_url":null},{"position":24,"raw":"[Muñoz et al., 2018] M. A. Muñoz, L. Villanova, D. Baatar, and K. Smith-Miles. Instance spaces for machine learning classification. Machine Learning, 107(1):109-147, 2018.","title":"Instance spaces for machine learning classification","authors":"M A Muñoz, L Villanova, D Baatar, K Smith-Miles","year":"2018","venue":"Machine Learning","doi":null,"url":null,"internal_url":null},{"position":25,"raw":"[Olson et al., 2016] R. S. Olson, N. Bartley, R. J. Urbanow- icz, and J. H. Moore. Evaluation of a Tree-based Pipeline Optimization Tool for Automating Data Science. In Proc. ACM-GECCO, pages 485-492. ACM Press, 2016.","title":"Evaluation of a Tree-based Pipeline Optimization Tool for Automating Data Science","authors":"R S Olson, N Bartley, R J Urbanowicz, J H Moore","year":"2016","venue":"Proc. ACM-GECCO","doi":null,"url":null,"internal_url":null},{"position":26,"raw":"[Pedregosa et al., 2011] F. Pedregosa, G. Varoquaux, A. Gramfort, et al. Scikit-learn: Machine learning in Python. JMLR, 12:2825-2830, 2011.","title":"Scikit-learn: Machine learning in Python","authors":"F Pedregosa, G Varoquaux, A Gramfort","year":"2011","venue":"JMLR","doi":null,"url":null,"internal_url":null},{"position":27,"raw":"[Silver et al., 2017] D. Silver, J. Schrittwieser, et al. Master- ing the Game of GO without Human Knowledge. Nature, 550(7676):354-359, 2017.","title":"Mastering the Game of GO without Human Knowledge","authors":"D Silver, J Schrittwieser","year":"2017","venue":"Nature","doi":null,"url":null,"internal_url":null},{"position":28,"raw":"[Snoek et al., 2015] J. Snoek, O. Rippel, K. Swersky, et al. Scalable Bayesian Optimization using Deep Neural Net- works. In Proc. ICML, pages 2171-2180, 2015.","title":"Scalable Bayesian Optimization using Deep Neural Networks","authors":"J Snoek, O Rippel, K Swersky","year":"2015","venue":"Proc. ICML","doi":null,"url":null,"internal_url":null},{"position":29,"raw":"[Swersky et al., 2014] K. Swersky, J. Snoek, and R. P. Adams. Freeze-thaw Bayesian optimization. preprint arXiv:1406.3896, 2014.","title":"Freeze-thaw Bayesian optimization","authors":"K Swersky, J Snoek, R P Adams","year":"2014","venue":null,"doi":null,"url":null,"internal_url":null},{"position":30,"raw":"[Vanschoren et al., 2013] J. Vanschoren, J.N. van Rijn, B. Bischl, and L. Torgo. OpenML: Networked Science in Machine Learning. SIGKDD Expl., 15(2):49-60, 2013.","title":"OpenML: Networked Science in Machine Learning","authors":"J Vanschoren, J N Van Rijn, B Bischl, L Torgo","year":"2013","venue":"SIGKDD Expl","doi":null,"url":null,"internal_url":null},{"position":31,"raw":"[Wang, 2016] Z. Wang. Practical and Theoretical Advances in Bayesian Optimization. PhD thesis, Univ. Oxford, 2016.","title":"Practical and Theoretical Advances in Bayesian Optimization","authors":"Z Wang","year":"2016","venue":null,"doi":null,"url":null,"internal_url":null},{"position":32,"raw":"[Wistuba, 2018] M. Wistuba. Deep Learning Architecture Search by Neuro-Cell-Based Evolution with Function- Preserving Mutations. In Berlingerio, M. et al., editor, Proc. ECML-PKDD, pages 243-258. Springer, 2018.","title":"Deep Learning Architecture Search by Neuro-Cell-Based Evolution with Function-Preserving Mutations","authors":"M Wistuba","year":"2018","venue":"Proc. ECML-PKDD","doi":null,"url":null,"internal_url":null},{"position":33,"raw":"[Wolpert, 1996] D. H. Wolpert. The Lack of A Priori Dis- tinctions Between Learning Algorithms. Neural Compu- tation, 8(7):1341-1390, 1996.","title":"The Lack of A Priori Distinctions Between Learning Algorithms","authors":"D H Wolpert","year":"1996","venue":"Neural Computation","doi":null,"url":null,"internal_url":null}],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}