结合先验知识的Bayesian优化算法研究与仿真

Improved Bayesian Optimization Algorithm Incorporated Prior Knowledge

由于一般优化问题的先验知识很难获取,因此在Bayesian网络学习中结合与利用先验知识一直是一个很难突破的问题。针对Bayesian优化算法(BOA)的特点,对一般优化问题如何发现和利用先验知识进行了分析讨论,把BOA中前一代种群所提供的信息作为先验知识结合到当前代Bayesian网络的学习中,提高了所学习网络的可靠性,从而提高算法的性能。仿真结果表明所提算法比传统BOA具有更强的全局寻优能力。

Since it is difficult to obtain the prior knowledge of general optimization problems,many algorithms of learning Bayesian networks almost do not incorporate or use the prior knowledge of problems. And it is NP-hard to learn the Bayesian networks. According to the characteristics of the Bayesian optimization algorithm （BOA）,it was discussed on how to discovery and use prior knowledge in general optimization problems. An improved Bayesian optimization algorithm incorporated prior knowledge was proposed. The information provided by the previous generation was considered as prior knowledge to be incorporated in the Bayesian networks learning. So the reliability of the networks and the performance of the proposed algorithm were improved. Simulation results show that the proposed algorithm achieves a stronger ability in searching the global optima than those of traditional BOA.