Historique: Machine Learning, Michèle Sebag et Francois Landes

Source de la version: 62 (courant)

''Exam: text in English; documents allowed; Internet not allowed.'' Past exams: see the Google Drive for the subjects in English ! https://drive.google.com/drive/u/1/folders/1uOfxeRUkGlGMuRlQchjpqMsuTUi4HxQN -= October 2nd, 2019 =- # Decision Trees ## Slides: {file name="DecisionTree_2019.pdf"} # Validation ## Slides: {file name="main_Validation_2019.pdf"} -= October 9th =- # Support Vector Machines ## Slides: {file name="Slides_SVM_2019.pdf"} ! Articles for seminar courses You can now pick your time slot: [https://docs.google.com/spreadsheets/u/1/d/1juj6wrUNKsaKWm028XLiwyleE2rdHXUFJAHyoYtCUWU/edit#gid=0|here] ''Select a paper in pairs or solo (this you are supposed to have done already)'' ''Prepare a written summary (2 pages pdf) - you will send it to us before'' ''Prepare an oral (~10-15 slides is advised, 15 min presentation + 5 min questions = 20 min total time).'' ''Papers in each bloc are related (they cite each other, improve on each other): discuss with the other pairs in order not to repeat same '' ''things and make the presentations most interesting for the audience (you can share your summaries to know what each other is doing)''. ''When the link is not there, google the title.'' ''Names in red are the names of students who selected a paper (therefore, it's no longer "available"). To choose a paper, send us an email. '' -= ~~#009:Papers added, Oct. 2019~~ =- * [http://www.ecmlpkdd2018.org/wp-content/uploads/2018/09/102.pdf|Multiple Instance Learning with Bag-Level Randomized Trees], ECML PKDD 2018, ~~#F00:Corentin Leloup and Maxime Chor~~ * [http://www.ecmlpkdd2018.org/wp-content/uploads/2018/09/125.pdf|One-Class Quantification], ECML PKDD 2018 * [https://www.researchgate.net/publication/324587839_Visualizing_the_Feature_Importance_for_Black_Box_Models|Visualizing the Feature Importance for Black Box Models], ECML PKDD 2018 * [http://www.ecmlpkdd2018.org/wp-content/uploads/2018/09/250.pdf|Efficient Estimation of AUC in a Sliding Window], ECML PKDD 2018 * [http://www.ecmlpkdd2018.org/wp-content/uploads/2018/09/578.pdf|Evaluation Procedures for Forecasting with Spatio-Temporal Data], ECML PKDD 2018 * [http://www.ecmlpkdd2018.org/wp-content/uploads/2018/09/596.pdf|Machine Learning for Targeted Assimilation of Satellite Data], ECML PKDD 2018 ~~ #F00:Chhivchung LAY ~~ * [http://www.marc-boulle.fr/publications/BoulleECML14.pdf|Towards Automatic Feature Construction for Supervised Classification], ECML PKDD 2014 * [https://hal.inria.fr/hal-01109774/document|Combination of One-Class Support Vector Machines for Classification with Reject Option], ECML PKDD 2014 * [https://link.springer.com/article/10.1007/s10994-015-5507-y|A Decomposition of the Outlier Detection Problem into a Set of Supervised Learning Problems ], ECML PKDD 2015 ~~#F00:Amritansh and Kunal~~ * [https://link.springer.com/chapter/10.1007/978-3-319-23528-8_1|Data split strategies for evolving predictive models ], ECML PKDD 2015 -= First papers =- !! EM and Bayes # [Naive Bayes models for probability estimation], ICML 2005 ~~#F00:LOPEZ ESTIGARRIBIA, Pablo Jose~~ !! Decision trees # [https://homes.cs.washington.edu/~pedrod/papers/kdd00.pdf|Mining High-Speed Data Streams] MLDM 2000 ~~#F00:Robin de Groot, Jérémy Hutin~~ # [https://link.springer.com/content/pdf/10.1007%2F11564126_50.pdf|Stress-Testing Hoeffding Trees] PKDD 2005 ~~ #F00:Sara Días, Ricardo Rojas~~ # [https://core.ac.uk/download/pdf/151040580.pdf|Extremely Fast Decision Tree Mining for Evolving Data Streams], KDD 2017 ~~ #F00:René Gómez , Ainhoa Zapirain ~~ # [https://www.mlpack.org/papers/ds.pdf|Very Simple Classification Rules Perform Well on Most Commonly Used Datasets], MLJ 93, # [https://arxiv.org/pdf/1403.3378.pdf|Box Drawings for Learning with Imbalanced Data], KDD 2014 ~~#F00:Pratham Solanki,Edoardo Conte~~ # [http://www.jmlr.org/papers/volume18/17-716/17-716.pdf|Learning Certifiably Optimal Rule Lists for Categorical Data], JMLR 2018 # [https://link.springer.com/article/10.1023%2FA%3A1010933404324|Random Forests], MLJ 2001, ~~#F00: LUO Hongyi LI Ziheng~~ # [http://proceedings.mlr.press/v37/nan15.html|Feature-Budgeted Random Forest], ICML 2015, ~~#F00:Amine Kili~~ !!! Applications of decision trees. # [https://www.sciencedirect.com/science/article/pii/S1746809417300204?via%3Dihub|Decision tree and random forest models for outcome prediction in antibody incompatible kidney transplantation], Biomedical signal processing and control, 2019 ~~#F00:Gabriela Martinez Lopera,Jose Badillo Puebla~~ # [Using Cutting-Edge Tree-Based Stochastic Models to Predict Credit Risk], 2018 (google the title); ~~#F00:Braulio Blanco Lambruschini, Tzu-Man Wu~~ # [http://isyou.info/inpra/papers/inpra-v5n4-02.pdf|Machine Learning based Approach to Financial Fraud Detection...], 2017; ~~#F00:Buse Ozer, Eugen Patrascu ~~ !! Estimation for inference and ensembles of classifiers # [https://icml.cc/Conferences/2005/proceedings/papers/027_Experimental_EspositoSaitta.pdf|Experimental comparison between bagging and Monte Carlo ensemble classification], ICML 2005 ~~ #F00:Gaspard Donada Vidal~~ # [https://link.springer.com/content/pdf/10.1007%2F978-3-642-04174-7_17.pdf|Classifier Chains for Multi-label Classification], ECML PKDD 2009 ~~ #F00:Timothée Babinet~~ # [An Analysis of Chaining in Multi-Label Classification], ECAI 2012 (google the title) ~~ #F00:Nicolas Devatine & Xinneng Xu~~ !! Area under the ROC curve # [http://dmip.webs.upv.es/ROCAI2004/papers/11-ecai2004-TobiasSing-A4.pdf|Learning Mixtures of Localized Rules by Maximizing the Area Under the ROC Curve], ECAI 2004 ~~ #F00:Sebastien Warichet,Theo Deschamps-Berger ~~ # [https://pdfs.semanticscholar.org/9211/b66736542c5dd88bc5694309619f97f07475.pdf|Learning Interestingness Measures in Terminology Extraction: A ROC-based approach], ICDM 2004 ~~ #F00: Ramine HAMIDI, Florian Bertelli~~ # [https://www.tau.ac.il/~saharon/papers/auc-fixed.pdf| Model selection via the AUC], ICML 2004 # [Optimizing abstaining classifiers using ROC analysis], ICML 2005 !! Finding good hyper-parameters # [http://jmlr.csail.mit.edu/papers/volume13/bergstra12a/bergstra12a.pdf|Random search for hyper-parameter optimization], JMLR 2012, ~~#F00:Gaetano D'Agostino and Francesco Cian~~ # [https://www.cs.ubc.ca/~hutter/papers/10-TR-SMAC.pdf|Sequential model-based optimization for general algorithm configuration], LION 2011 # [http://proceedings.mlr.press/v28/bardenet13.pdf|Collaborative hyperparameter tuning], ICML 13 ~~ #F00:Clément Veyssière, Eric Wang~~ # [https://arxiv.org/pdf/1406.3896.pdf|FREEZE-THAW BAYESIAN OPTIMIZATION], ArXiv 1014 ~~#F00:Eva Agrawal, Wafa Bouzouita ~~ # [https://www.ijcai.org/Proceedings/15/Papers/487.pdf|Speeding Up Automatic Hyperparameter Optimization of Deep Neural Networks by Extrapolation of Learning Curves], IJCAI 2015 !! Support Vector Machines # [Using String Kernels to Identify Famous Performers from Their Playing Style], ECML PKDD 2004 ~~#F00:Woody HOJEILY ~~ # [https://www.cs.cornell.edu/people/tj/publications/finley_joachims_05a.pdf|Supervised clustering with support vector machines], ICML 2005, ~~#F00:Moez Ezzeddine et Ghassen Chaabane~~ # [Adapting two-class support vector classification methods to many class problems], ICML 2005 # [A support vector method for multivariate performance measures], ICML 2005 # [An efficient method for simplifying support vector machines], ICML 2005 -= =-