Classification Performance Comparison between RVM and SVM

Both relevant vector machine and support vector machine are newly promoted pattern recognition algorithms. An extensive relevant literature displays that they have become hot topics in the field of machine learning. Due to the difference of their mechanism, little research is done to compare their performance. This paper experimentally compared several features of RVM with SVM which can characterize the classification performance on the basis of deeply understanding their algorithms. The results show that RVM is almost equal to SVM on training efficiency and classification accuracy, but as to sparse property, generalization ability and decision speed, RVM performs better. So it is recommended to study RVM deeply and extend its application areas further.