Real-coded Estimation of Distribution Algorithm by Using Probabilistic Models with Multiple Learning Rates
Masahiro Nakao, Tomoyuki Hiroyasu, Mitsunori Miki, Hisatake Yokouchi, Masato Yoshimi · Procedia Computer Science · 2011
Here, a new Real-coded Estimation of Distribution Algorithm (EDA) is proposed. The proposed EDA is called Real-coded EDA using Multiple Probabilistic Models (RMM). RMM includes multiple types of probabilistic models with different learning rates and diversities. The search capability of RMM was examined through several types of continuous test function. The results indicated that the search capability of RMM is better than or equivalent to that of existing Real-coded EDAs. Since better searching points are distributed for other probabilistic models positively, RMM can discover the global optimum in the early stages of the search.