Deriving fuzzy discretization from interval discretization

We examine two methods for deriving fuzzy discretization from interval discretization for classification problems. One method uses linear membership functions with trapezoidal shape. The other extends them to piecewise linear functions. In both methods, we use a control parameter that specifies the overlap grade between adjacent membership functions. The overlap grade can be viewed as the fuzzification grade of interval discretization. Through computer simulations, we examine three interval discretization methods from which fuzzy discretization is derived: equal width intervals, equal frequency intervals, and minimum entropy intervals. We extract fuzzy rules for wine data and sonar data with many continuous attributes. Simulation results show that the classification ability of fuzzy rules is improved in some cases and degraded in other cases by the increase in the fuzzification grade.

[1]  Aiko M. Hormann,et al.  Programs for Machine Learning. Part I , 1962, Inf. Control..

[2]  J. Ross Quinlan,et al.  C4.5: Programs for Machine Learning , 1992 .

[3]  María José del Jesús,et al.  Genetic feature selection in a fuzzy rule-based classification system learning process for high-dimensional problems , 2001, Inf. Sci..

[4]  Tapio Elomaa,et al.  General and Efficient Multisplitting of Numerical Attributes , 1999, Machine Learning.

[5]  Hisao Ishibuchi,et al.  Comparison of heuristic rule weight specification methods , 2002, 2002 IEEE World Congress on Computational Intelligence. 2002 IEEE International Conference on Fuzzy Systems. FUZZ-IEEE'02. Proceedings (Cat. No.02CH37291).

[6]  Heikki Mannila,et al.  Fast Discovery of Association Rules , 1996, Advances in Knowledge Discovery and Data Mining.

[7]  Hisao Ishibuchi,et al.  Performance evaluation of fuzzy partitions with different fuzzification grades , 2002, 2002 IEEE World Congress on Computational Intelligence. 2002 IEEE International Conference on Fuzzy Systems. FUZZ-IEEE'02. Proceedings (Cat. No.02CH37291).

[8]  Keki B. Irani,et al.  Multi-interval discretization of continuos attributes as pre-processing for classi cation learning , 1993, IJCAI 1993.

[9]  Tzung-Pei Hong,et al.  Trade-off Between Computation Time and Number of Rules for Fuzzy Mining from Quantitative Data , 2001, Int. J. Uncertain. Fuzziness Knowl. Based Syst..

[10]  Hisao Ishibuchi,et al.  Effect of rule weights in fuzzy rule-based classification systems , 2000, Ninth IEEE International Conference on Fuzzy Systems. FUZZ- IEEE 2000 (Cat. No.00CH37063).

[11]  Hisao Ishibuchi,et al.  Fuzzy data mining: effect of fuzzy discretization , 2001, Proceedings 2001 IEEE International Conference on Data Mining.

[12]  J. Ross Quinlan,et al.  Improved Use of Continuous Attributes in C4.5 , 1996, J. Artif. Intell. Res..

[13]  H. Ishibuchi,et al.  Distributed representation of fuzzy rules and its application to pattern classification , 1992 .

[14]  Jose Jesus Castro-Schez,et al.  Use of a fuzzy machine learning technique in the knowledge acquisition process , 2001, Fuzzy Sets Syst..

[15]  Antonio González Muñoz,et al.  Including a simplicity criterion in the selection of the best rule in a genetic fuzzy learning algorithm , 2001, Fuzzy Sets Syst..