A new method is proposed to reduce outliers’ influence on OCSVM model. Ramp Loss function is introduced into OCSVM optimization. An iterative algorithm is proposed to solve this new OCSVM optimization. The outliers are first identified and then removed from the training set.