中文摘要：

从大型数据库中学习网络结构一直是贝叶斯网络学习的难点之一．针对此问题提出了一种混合算法．将粒子群优化法简单且全局寻优能力强的特点．以及遗传算法良好的并行计算能力进行有效的结合，以增加学习的精度和效率．最后以经典的Asia，Cancer网络为实例．并与文中算法进行比较，验证了该算法的有效性．

英文摘要：

Learning structure from large databases is one of the difficulties of learning Bayesian Networks. To cope with this problem, a new hybrid algorithm is proposed. By integrating PSO （particle swarm optimization） and GA effectively, it owns not only simply and strong global optimization of PSO, but also favorable parallel computing capability of GA. Therefore, the learning accuracy and efficiency can be increased. Finally the proposed algorithm is compared with other algorithms in typical Bayesian networks such as Asia and Cancer, experimental results show that the proposed algorithm is effective.