The main drawback of support vector machine (SVM) algorithm is that it needs large memory and long training time while handling large training data set. In order to speed up the training and improve classification accuracy, this paper proposes a binary classification model, which fuses the Bagging, SVM and Adaboost algorithm. And a kind of denoising algorithm is proposed. Contrast SVM, the SVM-Adaboost and classification model proposed in this paper by experiment. With the expanding of training data, this classification model has improved training speed significantly under the premise of improving accuracy.