一种混沌贝叶斯优化算法

为了减少贝叶斯优化算法的计算量,该文提出了一种混沌贝叶斯优化算法。用混沌随机序列产生贝叶斯优化算法的初始群体,利用混沌随机性、遍历性和对初始条件的敏感性的特点,提供给贝叶斯网络变量空间丰富的信息,有利于建立接近最优的贝叶斯网络。为增加群体的多样性同时减少贝叶斯网络的建立次数,采用混沌搜索方法对贝叶斯网络产生的新解进行变异寻优,以此为基础再建立贝叶斯网络。实验结果表明,与贝叶斯优化算法相比,混沌贝叶斯优化算法能有效减少计算量。

An Optimization Algorithm Combining Chaotic Sequences with Bayesian Optimization Algorithm

In the paper,to decrease the computational time,an optimization algorithm is proposed by combining chaotic sequences with Bayesian optimization algorithm (BOA).By using the chaotic properties of ergodicity,stochastic property,and sensitivity to the initial condition,the chaotic sequences are used to initialize the BOA's population to provide the Bayesian network with abundant information of the variable space,which is in favor of constructing a better Bayesian network.To increase the population diversity and reduce the construction times of the Bayesian network,chaos search is adopted to improve the solutions generated by the Bayesian network,and then the improved solutions are used to construct the next Bayesian network.The experimental results indicate that,by compared with the BOA,the proposed algorithm can reduce the computational time efficiently.