Publications

Publications and technical reports related to the project.

Journal papers

• J. Hernández-Orallo, P. Flach, C. Ferri "A Unified View of Performance Metrics: Translating Threshold Choice into Expected Classification Loss", Journal of Machine Learning Research, 13(Oct):2813−2869, 2012.
• J. Hernández-Orallo, P. Flach, C. Ferri "ROC Curves in Cost Space", Machine Learning Journal, Volume 93, Issue 1, pp 71-91, Springer, 2013.
• J. Hernández-Orallo "ROC Curves for Regression", Pattern Recognition, Volume 46, Issue 12, Pages 3395–3411, Elsevier, 2013.
• A. Bella, C. Ferri, J. Hernández-Orallo, M.J. Ramírez-Quintana “Aggregative Quantification for Regression”, Data Mining and Knowledge Discovery, Volume 28, Issue 2, pp 475-518, March 2014.
• J. Hernández-Orallo "Probabilistic reframing for cost-sensitive regression" (doi), ACM Transactions on Knowledge Discovery from Data, 8(4), ACM, 2014.
• C.F. Ahmed, N. Lachiche, C. Charnay, S. El Jelali, A. Braud "Flexible propositionalization of continuous attributes in relational data mining", Expert Systems with Applications (Elsevier), Vol. 42, Issue 21, Pages 7698–7709, 2015.
• J. Hernández-Orallo, C. Ferri, N. Lachiche, A. Martínez-Usó, M. J. Ramírez-Quintana "Binarised Regression Tasks: Methods and Evaluation Metrics" Data Mining and Knowledge discovery, to appear.
• N. Nikolaou, N. Edakunni, M. Kull, P. Flach, G. Brown "Cost-sensitive boosting algorithms: Do we really need them?", Machine Learning 104: 359. doi:10.1007/s10994-016-5572-x, 2016.

Conference papers

• C. Charnay, N. Lachiche, A. Braud "Incremental Construction of Complex Aggregates: Counting over a Secondary Table", International Conference on Inductive Logic Programming (ILP), Short Paper. Rio de Janeiro, Brasil, August 2013
• C. Charnay, N. Lachiche, A. Braud "Pairwise Optimization of Bayesian Classifiers for Multi-Class Cost-Sensitive Learning", IEEE International Conference on Tools with Artificial Intelligence (ICTAI). Washington D.C., USA, pp. 499--505, November 2013
• C. F. Ahmed, C. Charnay, N. Lachiche, A. Braud, "Reframing on Relational Data", International Conference on Inductive Logic Programming (ILP), Long Paper. Nancy, France, September 2014.
• C. Charnay, N. Lachiche, A. Braud Construction of Complex Aggregates with Random Restart Hill-Climbing, 24th International Conference on Inductive Logic Programming, ILP 2014, Revised Selected Papers, Nancy, France, pages 49--61, Jesse Davis and Jan Ramon (Eds.), Springer, Lecture Notes in Computer Science, Volume 9046, September 2014.
• Al-Otaibi, R., Kull, M., and Flach, P. (2014). LaCova: A tree-based multilabel classifier using label covariance as splitting criterion. In Proceedings of the IEEE 13th international Conference on Machine Learning and Application (ICMLA-2014), Detroit, USA. IEEE.
• C. F. Ahmed, N. Lachiche, C. Charnay, and A. Braud, "Reframing continuous input attributes", In Proceedings of the 26th IEEE International Conference on Tools with Artificial Intelligence (ICTAI), pp. 31-38, Limassol, Cyprus, November, 2014.
• M. Kull, P. A. Flach: "Reliability Maps: A Tool to Enhance Probability Estimates and Improve Classification Accuracy", ECML/PKDD (2) 2014: 18-33
• L. A. C. Millard, M. Kull, P. A. Flach: "Rate-Oriented Point-Wise Confidence Bounds for ROC Curves", ECML/PKDD (2) 2014: 404-421
• P. Flach, M. Kull: "Precision-Recall-Gain curves: PR analysis done right", In Advances in Neural Information Processing Systems (NIPS), pp. 838-846, 2015.
• R. Al-Otaibi, R. Prudencio, M. Kull, P. A. Flach: "Versatile Decision Trees for Learning Over Multiple Contexts", ECML/PKDD (1) 2015: 184-199
• M. Kull, P. Flach: "Novel Decompositions of Proper Scoring Rules for Classification: Score Adjustment as Precursor to Calibration", ECML/PKDD (1) 2015: 68-85
• A. Martínez-Usó, J. Hernández-Orallo: "Multidimensional prediction models when the resolution context changes", ECML/PKDD (2) 2015: 509-524
• A. Martínez-Usó, J. Hernández-Orallo, M.J. Ramírez-Quintana, F. Martínez Plumed: "Pentaho + R: an Integral View for Multidimensional Prediction Models", CAEPIA Conference 2015, Albacete, Spain, November 9-12, 2015, LNCS 9422.
• C. Ahmed, M. Samiullah, N. Lachiche, M. Kull, P. Flach: Reframing in Frequent Pattern Mining, IEEE 27th International Conference on Tools with Artificial Intelligence (ICTAI 2015), Vietri sul Mare, Italy, pages 799-806, November 2015.
• C. Charnay, N. Lachiche, A. Braud "CARAF: Complex Aggregates within Random Forests", 25th International Conference on Inductive Logic Programming (ILP'15), to appear.

Workshop papers

• C. F. Ahmed, N. Lachiche, C. Charnay, A. Braud, "Dataset Shift in a Real-Life Dataset", ECML-PKDD Workshop (LMCE), Regular Paper. Nancy, France, September 2014.
• R. Al-Otaibi, P.A. Flach, and M. Kull(2014). Multi-label classification: A comparative study on threshold selection methods. In First International Workshop on Learning over Multiple Contexts (LMCE) at ECML-PKDD 2014, Nancy, France.
• Kull, M. and Flach, P. (2014). "Patterns of dataset shift". In First International Workshop on Learning over Multiple Contexts (LMCE) at ECML-PKDD 2014, Nancy, France.
• J. Hernández-Orallo, N. Lachiche, A. Martínez-Usó "Predictive models for multidimensional data when the resolution context changes", LMCE 2014.
• C. Ferri, J. Hernández-Orallo, A. Martínez-Usó, M.J. Ramírez-Quintana "Identifying dominant models when the noise context is known", LMCE 2014.
• M. Samiullah , C. F. Ahmed , A. Fariha , A. U. Ahmed, "Efficient Graph Classification in Shifted Datasets Using Weighted Correlated Feature Selection", ECML-PKDD Workshop (LMCE), Nancy, France, September 2014.
• A. Martínez-Usó, J. Hernández-Orallo, C. Ferri, M.J. Ramírez-Quintana. "Ordinal model reuse and selection for a varying number of categories", ECML-PKDD Workshop (LMCE), Porto, Portugal, September 2015.
• C. F. Ahmed. Reframing Bike Challenge Problem using Model Selection, ECML-PKDD Discovery Challenge MoReBikeS, Porto, Portugal, September 2015.
• Md. G. Rahman. An Introductory Survey on Reframing in Clustering, ECML-PKDD Workshop (LMCE), Porto, Portugal, September 2015.
• M. Kull, C. Ferri, A. Martínez-Usó. "Bike Rental and Weather Data across Dozens of Cities", ICML 2015 Workshop on Demand Forecasting, 2015.
• M. Kull, N. Lachiche, A. Martínez-Usó. "Model Reuse with Bike rental Station data", Proceedings of the ECML/PKDD 2015 Discovery Challenges, CEUR-WS.org, online CEUR-WS.org/Vol-1526/bikes_preamble.pdf, December 2015
• R. Al-Otaibi, M. Kull and P.A. Flach. "Multi-label Classification by Label Clustering based on Covariance". Proceedings of the ECML PKDD PhD Consortium 2015, Porto, Portugal.
• P. Flach. Cost-Sensitive Classification Meets Proper Scoring Rules. In 2nd International Workshop on Learning over Multiple Contexts (LMCE) at ECML-PKDD 2015, Porto, Portugal, September 2015.
• C. Ferri, M.J. Ramírez-Quintana, M. Kull, A. Martínez-Usó, P. Flach. An adaptive probabilistic classification method for dynamic class hierarchies. In 2nd International Workshop on Learning over Multiple Contexts (LMCE) at ECML-PKDD 2015, Porto, Portugal, September 2015.
• R. Prudencio, J. Hernández-Orallo, A. Martínez-Usó. Analysis of instance hardness in machine learning using item response theory. In 2nd International Workshop on Learning over Multiple Contexts (LMCE) at ECML-PKDD 2015, Porto, Portugal, September 2015.

Technical reports

Glossary

Software

Some software packages have been developed for the project and are publicly available:

• Rate Driven Curves: ROC Curves in Cost Space
• Brier Curves: a New Cost-Based Visualisation of Classifier Performance
• ROC analysis for regression
• Regression to Classification Curves Software
• SBA: Calibration techniques for multi-class problems
• Resolution context changes in multidimensional data prediction
• Historical bike sharing data from Valencia can be found at BitBucket account.
• A complete library in R for creating plots and metrics (UCE and OCE curves) is available at Regr2Clas library.
• Precision-Recall-Gain curves Software for visualising Precision-Recall-Gain curves and calculating the area under the curve, libraries available for R and Matlab.
• Integration of context aware widgets in the clowdflows.org network of online data mining services and a new node, clowdflows.unistra.fr, set up in Strasbourg.