ABSTRACT
This chapter presents various feature selection methods, classification methods, fitness evaluation methods, and ensemble techniques. Since the success of a genetic algorithm or genetic programming depends on a number of factors, including the fitness function used to evaluate the fitness of an individual, we present various fitness evaluation techniques for classification of balanced and unbalanced data. In many classification problems, single rules of genetic programming do not produce good test accuracy. To improve the performance of genetic programming, the majority voting genetic programming classifier (MVGPC) has been presented. The effectiveness of MVGPC is demonstrated by applying it to the classification of microarray data and unbalanced data. Though adaptive boosting (AdaBoost) is widely used as an accuracy improvement ensemble technique, we show that MVGPC performs better than AdaBoost for certain types of problems, such as classification of microarray data and unbalanced data. Finally, for an ensemble technique, various performance improvement techniques, such as diversity measurement and reduction of execution time, are discussed in detail.
