中文摘要：

支持向量机是一种基于结构风险最小化原理的学习技术，也是一种新的具有很好泛化性能的回归方法。目前，如何设计快速有效的回归估计算法仍然是支持向量机实际应用中的问题之一。文中对标准SVM回归估计算法加以改进，提出一种改进的SVM回归估计算法，并从学习速度和回归估计精度两个方面对提出的改进的SVM回归估计算法与标准SVM回归估计算法进行了比较。实验结果表明，在学习速度与回归估计精度之间取折衷时，文中提出的回归估计算法自由度更大。

英文摘要：

Support vector machine is a learning technique based on the structural risk minimization principle as well as a new regression method with good generalization ability. Now, how to design fast and efficient SVM algorithms applied to regression estimation becomes a great challenge in practical applications of support vector machine. Based on the normal support vector machine for regression estimation, an improved regression estimation algorithm of SVM is presented in this paper. Then,comparision of the proposed algorithm and the norreal regression estimation algorithm is implemented in learning speed and regression estimation precision. The experimental results show that the improved algorithm is better than the normal regression estimation algorithm when learning speed and regression estimation precision are considered.