Learning k-maxitive fuzzy measures from data by mixed integer programming

Abstract Fuzzy measures model interactions between the inputs in aggregation problems. Their complexity grows exponentially with the dimensionality of the problem, and elicitation of fuzzy measure coefficients either from domain experts or from empirical data poses a significant challenge. The notions of k-additivity and k-maxitivity simplify the fuzzy measures by limiting interactions to subsets of up to k elements. Learning fuzzy measures from data is an important elicitation technique which relies on solving an optimisation problem. A heuristic learning algorithm to identify k-maxitive fuzzy measures from the data on the basis of HLMS (Heuristic Least Mean Squares) was recently presented in Murillo et al. (2017) [11] . We present an alternative formulation of the fitting problem which delivers a globally optimal solution through the solution of a mixed integer programming (MIP) problem. To deal with high computational cost of MIP in moderate to large dimensions, we also propose a simple MIP relaxation technique which involves solving two related linear programming problems. We also provide a linear programming formulation for fitting k-tolerant fuzzy measures. We discuss implementations of the fitting methods and present the results of numerical experiments.

[1]  Christophe Labreuche,et al.  To : Fuzzy Sets and Systems , 2010 .

[2]  Didier Dubois,et al.  The Use of the Discrete Sugeno Integral in Decision-Making: A Survey , 2001, Int. J. Uncertain. Fuzziness Knowl. Based Syst..

[3]  Gang Li,et al.  Learning Choquet-Integral-Based Metrics for Semisupervised Clustering , 2011, IEEE Transactions on Fuzzy Systems.

[4]  Jian-Zhang Wu,et al.  Compromise principle based methods of identifying capacities in the framework of multicriteria decision analysis , 2014, Fuzzy Sets Syst..

[5]  L. Shapley.  A Value for n-person Games , 1988 .

[6]  Serge Guillaume,et al.  k-maxitive fuzzy measures: A scalable approach to model interactions , 2017, Fuzzy Sets Syst..

[7]  Michel Grabisch,et al.  A new algorithm for identifying fuzzy measures and its application to pattern recognition , 1995, Proceedings of 1995 IEEE International Conference on Fuzzy Systems..

[8]  Endre Pap,et al.  Two kinds of explicit preference information oriented capacity identification methods in the context of multicriteria decision analysis , 2018, Int. Trans. Oper. Res..

[9]  Hans-Jürgen Zimmermann,et al.  Improved feature selection and classification by the 2-additive fuzzy measure , 1999, Fuzzy Sets Syst..

[10]  Jean-Luc Marichal,et al.  Determination of weights of interacting criteria from a reference set , 2000, Eur. J. Oper. Res..

[11]  菅野 道夫,et al.  Theory of fuzzy integrals and its applications , 1975 .

[12]  Serge Guillaume,et al.  Revised HLMS: A useful algorithm for fuzzy measure identification , 2013, Inf. Fusion.

[13]  Bernard De Baets,et al.  Aggregation Operators Defined by k-Order Additive/Maxitive Fuzzy Measures , 1998, Int. J. Uncertain. Fuzziness Knowl. Based Syst..

[14]  Michel Grabisch,et al.  K-order Additive Discrete Fuzzy Measures and Their Representation , 1997, Fuzzy Sets Syst..

[15]  Michel Grabisch,et al.  The representation of importance and interaction of features by fuzzy measures , 1996, Pattern Recognit. Lett..

[16]  Jean-Luc Marichal,et al.  k-intolerant capacities and Choquet integrals , 2007, Eur. J. Oper. Res..

[17]  Michel Grabisch,et al.  A review of methods for capacity identification in Choquet integral based multi-attribute utility theory: Applications of the Kappalab R package , 2008, Eur. J. Oper. Res..

[18]  Jean-Luc Marichal,et al.  Tolerant or intolerant character of interacting criteria in aggregation by the Choquet integral , 2004, Eur. J. Oper. Res..

[19]  Gleb Beliakov,et al.  Nonmodularity index for capacity identifying with multiple criteria preference information , 2019, Inf. Sci..

[20]  Humberto Bustince,et al.  A Practical Guide to Averaging Functions , 2015, Studies in Fuzziness and Soft Computing.

[21]  Liping Yu,et al.  Using the monotone measure sum to enrich the measurement of the interaction of multiple decision criteria , 2016, J. Intell. Fuzzy Syst..

[22]  Alain Chateauneuf,et al.  Some Characterizations of Lower Probabilities and Other Monotone Capacities through the use of Möbius Inversion , 1989, Classic Works of the Dempster-Shafer Theory of Belief Functions.

[23]  Radko Mesiar,et al.  Generalizations of k-order additive discrete fuzzy measures , 1999, Fuzzy Sets Syst..

[24]  M. Grabisch.  The application of fuzzy integrals in multicriteria decision making , 1996 .

[25]  Gleb Beliakov,et al.  Construction of aggregation functions from data using linear programming , 2009, Fuzzy Sets Syst..

[26]  Gleb Beliakov,et al.  Nonadditivity index and capacity identification method in the context of multicriteria decision making , 2018, Inf. Sci..

[27]  Shanlin Yang,et al.  2-Additive Capacity Identification Methods From Multicriteria Correlation Preference Information , 2015, IEEE Transactions on Fuzzy Systems.

[28]  Liping Yu,et al.  The sum interaction indices of some particular families of monotone measures , 2016, J. Intell. Fuzzy Syst..

[29]  Gleb Beliakov,et al.  Aggregation Functions: A Guide for Practitioners , 2007, Studies in Fuzziness and Soft Computing.

[30]  Gleb Beliakov,et al.  Probabilistic bipartition interaction index of multiple decision criteria associated with the nonadditivity of fuzzy measures , 2018, Int. J. Intell. Syst..

[31]  Radko Mesiar,et al.  K-Order Additive Fuzzy Measures , 1999, Int. J. Uncertain. Fuzziness Knowl. Based Syst..