A new resampling method of imbalanced large data based on class boundary

We propose a new method for the calculation of class boundary. Through the compression of large data sets, the method can remove the samples which are not in the class boundary and have little effect on the classification results. It can also improve the classification accuracy of the traditional algorithm by selecting appropriate threshold. For the imbalanced data sets, this method can remove the samples of majority class on the class boundary and improve the classification performance of the minority class. The experiment has demonstrated that the method is effective.