Abstract: This paper combines Support Vector Machine（SVM） with semi-supervised theory，brings out a semi-supervised regression model based on SVM co-training，using two support vector regressors co-training.Using experimental datasets to compare with supervised SVM model and semi-supervised self-training model，the experimental results show that semi-supervised regression model based on SVM co-training can improve regression estimates when lack of labeled samples.

Key words: semi-supervised learning, Support Vector Machine（SVM）, co-training, self-training

摘要： 将支持向量机与半监督学习理论相结合，提出基于支持向量机协同训练的半监督回归模型，使用两个支持向量机回归模型相互影响，协同训练。利用实验数据集进行实验，并与监督支持向量机回归模型、半监督自训练支持向量机回归模型作比较。实验结果表明，基于支持向量机协同训练的半监督回归模型在缺少标记样本的情况下，提高了回归估计的精度。

关键词: 半监督学习, 支持向量机, 协同训练, 自训练