A Novel Measure of Diversity for Support Vector Machine Ensemble

The diversity of an ensemble is deemed to be a key factor which determines performance in ensemble learning. A variety of approaches have been advanced to quantify diversity by analyzing the prediction of classification which relies on the validation set. This paper proposes a new method how to measure diversity and ensemble for linear kernel Support Vector Machine, which is based on the characteristic parameters of Support Vector Machine. The new method is proved to achieve better performance than the traditional measures of diversity such as Discrepancy method. Further research on relationship between diversity and accuracy is conducted by the method.