Performance-Improved AdaBoost with Parameterized and Simple-Operation Genetic Algorithm

In this paper, we propose a design methodology for the AdaBoost with parameterized genetic algorithm (PGA) to improve the accuracy performance. It can provide 3 parameters to arbitrarily adjust. In addition, we develop the low-computing operations for crossover and mutation tasks, the essential processing in GA. In the performance analysis, our 6classifier AdaBoost with PGA method outperforms 10-classifier AdaBoost in terms of accuracy. While the number of classifiers survived is 2, 3, and 4, the according accuracy has a maximum increase of 5.74%, 7.65%, and 8.81%, respectively. Its benefits include a smaller number of classifiers required and accuracy increase.