交互支持向量机学习算法及其应用

Interactive support vector machine learning algorithm and its application

交互支持向量机学习算法能解决一些监督学习问题中学习样本较少的问题，它以支持向量机（ Ｓ Ｖ Ｍ ）方法为基础，将设计分类器变成一个交互的过程，即： 根据对已知样本进行的 Ｓ Ｖ Ｍ 分类器设计，主动采样选择“有用”的新样本，并进行下一步 Ｓ Ｖ Ｍ 分类器的设计。与普通 Ｓ Ｖ Ｍ 法相比，该方法所需的样本量大大降低，而且可能达到更好的推广能力。文本信息过滤问题的实例说明了该算法的有效性。

Interactive support vector machine (SVM) can solve some supervised learning problems which are lack of learning samples. The algorithm bases on SVM. An interactive process is introduced in classifier's design. According to the prior SVM classifier, an “useful” sample can be select actively, then you can get the better SVM calssifier. Comparing with the general SVM, it greatly reduces the number of samples, and probably reaches higher generalization ability. The application on text information filtering shows a successful example using this algorithm.