改进遗传-狼群对节点序寻优的贝叶斯网络结构算法

贝叶斯网络是数据挖掘领域的一种重要方法。针对贝叶斯网络结构学习算法寻优效率低和易陷入局部最优的问题，提出一种基于改进的混合遗传-狼群对节点序寻优的贝叶斯网络结构学习算法。该算法首先利用深度优先搜索对最大支撑树的节点进行拓扑排序；然后利用动态变异及最优交叉算子构建适用于节点序寻优的改进捕食行为，引入动态参数因子来增强算法局部寻优能力；最后与K2算法结合得到最优的贝叶斯网络结构。用3种不同大小的标准网络数据集中进行实验，结果表明，该算法收敛到较优值，寻优效率高于其它同类优化算法。

Bayesian Network Structure Learning for Node Order Optimization Based on Improved Genetic-Wolf Pack Algorithm

Bayesian network is an important method in the field of data mining. The Bayesian network structure learning algorithm is easy to fall into the problem of local optimization and low efficiency. A Bayesian network structure learning algorithm based on improved hybrid genetic wolves group is proposed to optimize the node order. Firstly, the algorithm uses the depth-first search to rank the nodes of the largest supporting tree. Then, using dynamic mutation and optimal crossover operator to construct predator behavior that suitable for node order optimization. The algorithm introduces dynamic parameter factors to enhance the ability of local optimization. Finally, the optimal Bayesian network structure is obtained by combining with K2 algorithm. Experiments are performed on three different sizes of standard network data sets. The simulation results show that the algorithm has high optimization and the optimization efficiency is higher than other similar optimization algorithms.