Study on the Performance of Decision Graph Bayesian Optimization Algorithm
Zhifu Shi, Haiyan Liu, Yin-sheng Zhu · 2009
The evolutions computation is the best proceeding algorithm for all kinds of optimization problem in the world. Bayesian optimization algorithm (BOA) is one kind of the evolution algorithm which is advantage on others for high order, hierarchical and correlative on anther optimization problem. For improving the ability of the BOA, the decision graph was introduced to enhance the represent and learn of Bayesian network and compress the parameter saving. The optimization mechanism and the algorithm model were studied in detail. The evaluation indexes and test function were also built for validating the merits. The performance and efficiency of DBOA were analyzed with compare with BOA, basic genetic algorithm and binary particle swarm optimization. The test results showed that DBOA was effective for hierarchical decomposable function.