引入支持向量机回归,提出具有数据修补功能的贝叶斯网络参数学习算法.该算法利用贝叶斯网络各观测节点不同时刻下的观测信息,在无先验信息约束下,通过样本回归对缺失数据进行修复.在获得的完整数据基础上利用最大似然估计完成贝叶斯网络参数估计.仿真结果表明,在有数据缺失的小样本情况下,该参数学习方法与标准EM算法相比,能够有效的提高参数学习效率以及推理结果的精度. The regression of support vector machine (SVM) is introduced to propose a Bayesian network parameter learning algorithm with data patching function. This algorithm uses the observed information of the observing nodes at different moments under the Bayesian network, The data is repaired.Based on the obtained complete data, the maximum likelihood estimation is used to complete the Bayesian network parameter estimation.The simulation results show that compared with the standard EM algorithm, Effectively improve the efficiency of parameter learning and the accuracy of inference results.