A Survey on ROC-based Ordinal Regression
Inhaltsverzeichnis
Reference
Willem Waegeman, Bernard De Baets: A Survey on ROC-based Ordinal Regression. In: Fürnkranz, J. and Hüllermeier, E.: Preference Learning, 2011, 127-154.
DOI
http://dx.doi.org/10.1007/978-3-642-14125-6_7
Abstract
Ordinal regression can be seen as a special case of preference learning, in which the class labels corresponding with data instances can take values from an ordered finite set. In such a setting, the classes usually have a linguistic interpretation attached by humans to subdivide the data into a number of preference bins. In this chapter, we give a general survey on ordinal regression from a machine learning point of view. In particular, we elaborate on some important connections with ROC analysis that have been introduced recently by the present authors. First, the important role of an underlying ranking function in ordinal regression models is discussed, as well as its impact on the performance evaluation of such models. Subsequently, we describe a new ROC-based performance measure that directly evaluates the underlying ranking function, and we place it in the more general context of ROC analysis as the volume under an r-dimensional ROC surface (VUS) for in general rclasses. Furthermore, we also discuss the scalability of this measure and show that it can be computed very efficiently for large samples. Finally, we present a kernel-based learning algorithm that optimizes VUS as a specific case of structured support vector machines.
Extended Abstract
Bibtex
@incollection{ year={2011}, isbn={978-3-642-14124-9}, booktitle={Preference Learning}, editor={Fürnkranz, Johannes and Hüllermeier, Eyke}, doi={10.1007/978-3-642-14125-6_7}, title={A Survey on ROC-based Ordinal Regression}, url={http://dx.doi.org/10.1007/978-3-642-14125-6_7, http://de.evo-art.org/index.php?title=A_Survey_on_ROC-based_Ordinal_Regression }, publisher={Springer Berlin Heidelberg}, author={Waegeman, Willem and Baets, BernardDe}, pages={127-154}, language={English} }
Used References
1. S.Agarwal, T.Graepel, R.Herbrich, S.Har-Peled, D.Roth, Generalization bounds for the area under the ROC curve. J. Mach. Learn. Res. 6, 393–425 (2005)MathSciNet 2. A.Agresti, Categorical Data Analysis, 2nd version. (Wiley, 2002) 3. K.Ataman, N.Street, Y.Zhang, Learning to rank by maximizing AUC with linear programming, in Proceedings of the IEEE International Joint Conference on Neural Networks(Vancouver, BC, Canada, 2006), pp. 123–129 4. C.J.C. Burges, A tutorial on support vector machines for pattern recognition. Data Min. Knowl. Discov. 2(2), 121–167 (1998)CrossRef 5. Z.Cao, T.Qin, T.Liu, M.Tsai, H.Li, Learning to rank: from pairwise approach to listwise approach, in Proceedings of the International Conference on Machine Learning(Corvallis, OR, USA, 2007), pp. 129–136 6. K.Cao-Van, Supervised Ranking, from semantics to algorithms. PhD thesis, Ghent University, Belgium, 2003 7. W.Chu, Z.Ghahramani, Gaussian processes for ordinal regression. J. Mach. Learn. Res. 6, 1019–1041 (2005)MathSciNetMATH 8. W.Chu, Z.Ghahramani, Preference learning with Gaussian processes, in Proceedings of the International Conference on Machine Learning(Bonn, Germany, 2005), pp. 137–144 9. W.Chu, S.Keerthi, New approaches to support vector ordinal regression, in Proceedings of the International Conference on Machine Learning(Bonn, Germany, 2005), pp. 321–328 10. W.Chu, S.Keerthi, Support vector ordinal regression. Neural Comput. 19(3), 792–815 (2007) 11. S.Clémençon, N.Vayatis, Ranking the best instances. J. Mach. Learn. Res. 8, 2671–2699 (2007)MathSciNetMATH 12. W.Cohen, R.Schapire, Y.Singer, Learning to order things, in Advances in Neural Information Processing Systems, vol.10 (MIT, Vancouver, Canada, 1998), pp. 451–457 13. C.Cortes, M.Mohri, AUC optimization versus error rate minimization, in Advances in Neural Information Processing Systems, vol.16 (MIT, Vancouver, Canada, 2003), pp. 313–320 14. C.Cortes, M.Mohri. Confidence intervals for the area under the ROC curve, in Advances in Neural Information Processing Systems, vol.17 (MIT, Vancouver, Canada, 2004), pp. 305–312 15. C.Cortes, M.Mohri, A.Rastogi, Magnitude-preserving ranking algorithms, in Proceedings of the International Conference on Machine Learning(Corvallis, OR, USA, 2007), pp. 169–176 16. K.Crammer, Y.Singer, Pranking with ranking, in Proceedings of the Conference on Neural Information Processing Systems(Vancouver, Canada, 2001), pp. 641–647 17. N.Cristianini, J.Shawe-Taylor, An Introduction to Support Vector Machines(Cambridge University Press, 2000) 18. B.De Baets, H.De Meyer, Transitivity frameworks for reciprocal relations: cycle-transitivity versus { FG}-transitivity. Fuzzy Sets Syst. 152, 249–270 (2005)MATHCrossRef 19. B.De Baets, H.De Meyer, B.De Schuymer, S.Jenei, Cyclic evaluation of transitivity of reciprocal relations. Soc. Choice Welfare 26, 217–238 (2006)MATHCrossRef 20. B.De Schuymer, H.De Meyer, B.De Baets, Cycle-transitive comparison of independent random variables. J. Multivar. Anal. 96, 352–373 (2005)MATHCrossRef 21. B.De Schuymer, H.De Meyer, B.De Baets, S.Jenei, On the cycle-transitivity of the dice model. Theory Decis. 54, 164–185 (2003)CrossRef 22. S.Dreiseitl, L.Ohno-Machado, M.Binder, Comparing three-class diagnostic tests by three-way ROC analysis. Med. Decis. Mak. 20, 323–331 (2000)CrossRef 23. E.Frank, M.Hall, A simple approach to ordinal classification, in Proceedings of the European Conference on Machine Learning, (Freibourg, Germany, 2001), pp. 146–156 24. Y.Freund, R.Yier, R.Schapire, Y.Singer, An efficient boosting algorithm for combining preferences. J. Mach. Learn. Res. 4, 933–969 (2003) 25. J.Fürnkranz, Round robin classification. J. Mach. Learn. Res. 2, 723–747 (2002) 26. J.Fürnkranz, E.Hüllermeier, Pairwise preference learning and ranking. Lect. Notes Comput. Sci. 2837, 145–156 (2003)CrossRef 27. M.Gönen, G.Heller, Concordance probability and discriminatory power in proportional hazards regression. Biometrika 92(4), 965–970 (2005)MathSciNetMATHCrossRef 28. D.Hand, R.Till, A simple generalization of the area under the ROC curve for multiple class problems. Mach. Learn. 45, 171–186 (2001)MATHCrossRef 29. J.Hanley, B.McNeil, The meaning and use of the area under a receiver operating characteristics curve. Radiology 143, 29–36 (1982) 30. J.Hanley, B.McNeil, A method of comparing receiver operating characteristics curves derived from the same class. Radiology 148, 839–843 (1983) 31. S.Har-Peled, D.Roth, D.Zimak, Constraint classification: A new approach to multi-class classification and ranking, in Advances in Neural Information Processing Systems, vol.15 (MIT, Vancouver, Canada, 2002), pp. 785–792 32. E.Harrington, Online ranking/collaborative filtering using the perceptron algorithm, in Proceedings of the 20th International Conference on Machine Learning(Washington, USA, 2003), pp. 250–257 33. T.Hastie, R.Tibshirani, Classification by pairwise coupling. Ann. Stat. 26(2), 451–471 (1998)MathSciNetMATH 34. M.Heller, B.Schnörr, Learning sparse representations by non-negative matrix factorization and sequential cone programming. J. Mach. Learn. Res. 7, 1385–1407 (2006)MathSciNet 35. R.Herbrich, T.Graepel, K.Obermayer, Large margin rank boundaries for ordinal regression, in Advances in Large Margin Classifiers, ed. by A.Smola, P.Bartlett, B.Schölkopf, D.Schuurmans (MIT, 2000), pp. 115–132 36. A.Herschtal, B.Raskutti, Optimizing area under the ROC curve using gradient descent, in Proceedings of the International Conference on Machine Learning(Bonn, Germany, 2005), pp. 49–57 37. J.Higgins, Introduction to Modern Nonparametric Statistics(Duxbury, 2004) 38. E.Hüllermeier, J.Hühn, Is an ordinal class structure useful in classifier learning? Int. J. Data Min. Model. Manag. 1(1), 45–67 (2009) 39. T.Joachims, A support vector method for multivariate performance measures, in Proceedings of the International Conference on Machine Learning(Bonn, Germany, 2005), pp. 377–384 40. J.Kelley, The cutting plane method for convex programs. J. Soc. Ind. Appl. Math. 9, 703–712 (1960)MathSciNet 41. S.Kramer, G.Widmer, B.Pfahringer, M.Degroeve, Prediction of ordinal classes using regression trees. Fundam. Informaticae 24, 1–15 (2000)MathSciNet 42. G.Lanckriet, N.Cristianini, P.Bartlett, L.El Gaoui, M.Jordan, Learning the kernel matrix with semidefinite programming. J. Mach. Learn. Res. 5, 27–72 (2004)MATH 43. L.Lehmann, Nonparametrics: Statistical Methods based on Ranks(Holden Day, 1975) 44. S.Lievens, B.De Baets, K.Cao-Van, A probabilistic framework for the design of instance-based supervised ranking algorithms in an ordinal setting. Ann. Oper. Res. 163, 115–142 (2008)MathSciNetMATHCrossRef 45. H.Lin, L.Li, Large-margin thresholded ensembles for ordinal regression: Theory and practice. Lect. Notes Comput. Sci. 4264, 319–333 (2006)CrossRef 46. R.Luce, P.Suppes, Handbook of Mathematical Psychology, chapter Preference, Utility and Subjective Probability (Wiley, 1965), pp. 249–410 47. P.McCullagh, Regression models for ordinal data. J. R. Stat. Soc. B 42(2), 109–142 (1980)MathSciNetMATH 48. C.Nakas, C.Yiannoutsos, Ordered multiple-class ROC analysis with continuous measurements. Stat. Med. 22, 3437–3449 (2004)CrossRef 49. M.Öztürk, A.Tsoukiàs, Ph. Vincke, Preference modelling, in Multiple Criteria Decision Analysis. State of the Art Surveys, ed. by J.Figueira, S.Greco, M.Ehrgott (Springer, 2005), pp. 27–71 50. J.Platt, N.Cristianini, J.Shawe-Taylor, Large margin DAGs for multiclass classification. Adv. Neural Process. Syst. 12, 547–553 (2000) 51. R.Potharst, J.C. Bioch, Decision trees for ordinal classification. Intell. Data Process. 4(2) (2000) 52. J.D. Rennie, N.Srebro, Loss functions for preference levels: Regression with discrete, ordered labels in Proceedings of the IJCAI Multidisciplinary Workshop on Advances in Preference Handling, (Edinburgh, Scotland, 2005), pp. 180–186 53. R.Rifkin, A.Klautau, In defense of one-versus-all classification. J. Mach. Learn. Res. 5, 101–143 (2004)MathSciNetMATH 54. B.Schölkopf, A.Smola. Learning with Kernels, Support Vector Machines, Regularisation, Optimization and Beyond(MIT, 2002) 55. A.Shashua, A.Levin, Ranking with large margin principle: Two approaches, in Advances in Neural Information Processing Systems, vol.16 (MIT, Vancouver, Canada, 2003), pp. 937–944 56. V.Torra, J.Domingo-Ferrer, J.M. Mateo-Sanz, M.Ng, Regression for ordinal variables without underlying continuous variables. Inf. Sci. 176, 465–476 (2006)MathSciNetCrossRef 57. Y.Tsochantaridis, T.Joachims, T.Hofmann, Y.Altun, Large margin methods for structured and independent output variables. J. Mach. Learn. Res. 6, 1453–1484 (2005)MathSciNetMATH 58. G.Tutz, K.Hechenbichler, Aggregating classifiers with ordinal response structure. J. Stat. Comput. Simul. 75(5), (2004) 59. W.Waegeman, B.De Baets, On the ERA ranking representability of multi-class classifiers. Artif. Intell. (2009) submitted 60. W.Waegeman, B.De Baets, L.Boullart, Learning a layered graph with a maximal number of paths connecting source and sink, in Proceedings of the ICML Workshop on Constrained Optimization and Structured Output Spaces(Corvallis, OR, USA, 2007) 61. W.Waegeman, B.De Baets, L.Boullart, Learning layered ranking functions with structured support vector machines. Neural Netw. 21(10), 1511–1523 (2008)MATHCrossRef 62. W.Waegeman, B.De Baets, L.Boullart, On the scalability of ordered multi-class ROC analysis. Comput. Stat. Data Anal. 52, 3371–3388 (2008)MATHCrossRef 63. W.Waegeman, B.De Baets, L.Boullart, ROC analysis in ordinal regression learning. Pattern Recognit. Lett. 29, 1–9 (2008)CrossRef 64. J.Xu, Y.Cao, H.Li, Y.Huang, Cost-sensitive learning of SVM for ranking, in Proceedings of the 17th European Conference on Machine Learning(Berlin, Germany, 2006), pp. 833–840 65. L.Yan, R.Dodier, M.Mozer, R.Wolniewiecz, Optimizing classifier performance via an approximation to the Wilcoxon-Mann-Whitney statistic, in Proceedings of the International Conference on Machine Learning(Washington, DC, USA, 2003), pp. 848–855 66. S.Yu, K.Yu, V.Tresp, H.Kriegel, Collaborative ordinal regression, in Proceedings of the International Conference on Machine Learning(Pittsburgh, PA, 2006), pp. 1089–1096