Improved algorithm for AdaBoost with SVM base classifiers

The relation between the performance of AdaBoost and the performance of base classifiers was analyzed, and the approach of improving the classification performance of AdaBoostSVM was studied There is inconsistency existed between the accuracy and diversity of base classifiers, and the inconsistency affect generalization performance of the algorithm. A new variable sigma-AdaBoostSVM was proposed by adjusting the kernel function parameter of the base classifier based on the distribution of training samples. The proposed algorithm improves the classification performance by making a balance between the accuracy and diversity of base classifiers. Experimental results indicate the effectiveness of the proposed algorithm.