Selecting linguistic classification rules by two-objective genetic algorithms

Shows how two-objective genetic algorithms can be applied to a rule selection problem of linguistic classification rules. First the authors briefly describe a generation method of linguistic classification rules from numerical data. Next the authors formulate a rule selection problem of linguistic classification rules. This problem has two objectives: to maximize the number of correctly classified training patterns and to minimize the number of selected rules. Then the authors propose a two-objective genetic algorithm for finding non-dominated solutions of the rule selection problem. Finally, the authors extend their two-objective genetic algorithm to a hybrid algorithm where a learning method is applied to each individual (i.e., each rule set) generated in the execution of the two-objective genetic algorithm.