A Comparative Study of Sampling Methods in Node Histogram Models with Probabilistic Model-Building Genetic Algorithms

In a previous paper we proposed a node histogram based sampling algorithm (NHBSA) and compared it with edge histogram based sampling algorithm (EHBSA). The results showed NHBSA outperforms EHBSA on the permutation problems where absolute position of each node in a string is related to its performance. However, we used only a limited variation of sampling methods for NHBSA. In this paper, we propose several variations of sampling methods for NHBSA and explore conditions for them to work well with NHBSA.