Abstract: Semi-supervised classification algorithm attempts to establish a set of recognition methods and criteria for specific unknown samples based on known samples.PTSVM（Progressive Transductive Support Vector Machine） is a semi-supervised classification algorithm based on SVM.In this paper，a semi-supervised classification algorithm based on the combination of the separation degree and support vector machine is devised，which uses the separation degree in Fisher criteria as metric and Fisher criteria as evaluation function.The experimental results show that this algorithm achieves a better learning result than PTSVM on time complexity and sample test precision.

Key words: semi-supervised classification, support vector machine, separation degree

摘要： 半监督分类算法试图根据已知样本对特定的未知样本建立一套进行识别的方法和准则。渐进直推式分类学习算法是一种基于SVM的半监督分类学习方法，在基于渐进直推式分类学习算法的基础上，利用Fisher准则中的样本离散度作为度量标准，采用Fisher准则函数作为评价函数，提出了一种基于离散度量和SVM相结合的半监督分类算法，在时间复杂度和样本测试精度上较PTSVM算法都取得了良好的学习效果。

关键词: 半监督分类, 支持向量机, 离散度量