Automatic generation of fuzzy classification systems using hyper-cone membership functions

In this paper, we propose automatic generation methods of fuzzy classification rules with the Genetic Algorithms (GAs) to obtain compact fuzzy systems. This time, we propose an approach of hyper-cone membership function to construct rules for the antecedent part. Then, this method is used to determine the location and shape of hyper-cone membership function in the antecedent part, output class and the number of necessary inputs of each rule by GAs. Also, using the rule addition method in GA process, compact fuzzy classification systems are obtained. Though the proposed methods are quite simple, the process of GAs on both methods presents a solution for solving two-objective optimization problems: increasing the numbers of correct pattern classification, while decreasing the rule and input numbers optimally. This method was applied to Wine data sets and Wisconsin Prognostic Breast Cancer (WPBC) data sets. Wine data sets consist of 13 inputs and three outputs, while WPBC data sets contain 33 inputs and two outputs.

[1]  Bart Kosko,et al.  Fuzzy function approximation with ellipsoidal rules , 1996, IEEE Trans. Syst. Man Cybern. Part B.

[2]  Hisao Ishibuchi,et al.  Evolutionary algorithms for constructing linguistic rule-based systems for high-dimensional pattern classification problems , 1998, 1998 IEEE International Conference on Evolutionary Computation Proceedings. IEEE World Congress on Computational Intelligence (Cat. No.98TH8360).

[3]  Hiroyuki Inoue,et al.  Fuzzy classifier system using hyper-cone membership functions and rule reduction techniques , 2001, 10th IEEE International Conference on Fuzzy Systems. (Cat. No.01CH37297).

[4]  Katsuari Kamei,et al.  An Application of Fuzzy Clustering to Controller Design , 1996 .

[5]  D. J. Newman,et al.  UCI Repository of Machine Learning Database , 1998 .

[6]  H. Ishibuchi,et al.  Adjusting fuzzy partitions by genetic algorithms and histograms for pattern classification problems , 1998, 1998 IEEE International Conference on Evolutionary Computation Proceedings. IEEE World Congress on Computational Intelligence (Cat. No.98TH8360).

[7]  Yoshiteru Nakamori,et al.  Fuzzy Rule Generation by Hyperellipsoidal Clustering , 1996 .

[8]  Hideo Tanaka,et al.  Construction of fuzzy classification systems with rectangular fuzzy rules using genetic algorithms , 1994, CVPR 1994.

[9]  T. Fukuda,et al.  RBF-fuzzy system with GA based unsupervised/supervised learning method , 1995, Proceedings of 1995 IEEE International Conference on Fuzzy Systems..

[10]  Yoshiki Uchikawa,et al.  On fuzzy modeling using fuzzy neural networks with the back-propagation algorithm , 1992, IEEE Trans. Neural Networks.

[11]  Hiroyuki Inoue,et al.  Automatic Generation of Fuzzy If-Then Rules using Genetic Algorithms and Hyper-Cone Membership Functions , 1996 .

[12]  Hiroyuki Inoue,et al.  Automatic generation of fuzzy rules using hyper-elliptic-cone membership functions by genetic algorithms , 1998, J. Intell. Fuzzy Syst..

[13]  M.A. Lee,et al.  Integrating design stage of fuzzy systems using genetic algorithms , 1993, [Proceedings 1993] Second IEEE International Conference on Fuzzy Systems.

[14]  Kenji Miyasaka,et al.  A fuzzy classifier system using hyper-cone membership functions and its application to inverted pendulum control , 2002, IEEE International Conference on Systems, Man and Cybernetics.

[15]  Isao Hayashi,et al.  Construction of fuzzy inference rules by NDF and NDFL , 1992, Int. J. Approx. Reason..