Non-balanced Binary Tree with Active Learning Support Vector Machine

In order to solve the inherent defects of traditional two—class classification support vector machine when the types of data is variety and many samples is without being labeled,an algorithm was proposed through a combination of active learning and non—balanced binary tree multi—class classification support vector machine. First,a non—balanced binary tree structure is constructed through the class distance,from easy to difficult in turn construct node,put most likely to separate the class on the root node,then,using active learning strategies the corresponding labels will be added for the selected samples,and automatically the labeled samples will be append to the training sample set.The experimental results show that the performance of the proposed algorithm is superior to the ordinary active learning support vector machine algorithm,by improving the classification accuracy,and greatly reducing running time of the algorithm.