New Class Recognition Based on Support Vector Data Description

In online classification tasks, such as credit evaluation, spam detection and intrusion detection, new class of patterns sometimes emerges. In order to adapt classifier to the change of distribution, recognition of new class becomes a key problem. To deal with this problem, this paper proposes a novel method based on support vector data description. This method detects and recognizes new class by the description of known classes. The experimental results show that the proposed method can recognize new class well, and on the basis of this technology online classifier is adapted to the change well