利用条件概率和gibbs抽样技术为分布估计算法构造 - 控制理论与应用.pdfVIP

30 3 Vol. 30 No. 3 2013 3 Control Theory Applications Mar. 2013 DOI: 10.7641/CTA.2013.20835 Gibbs , (, 100083) : , Gibbs , . , , Gibbs . , , . : ; Gibbs; ; : TP181 : A General stochastic model for algorithm of distribution estimation with conditional probabilities and Gibbs sampling ZHANG Fang , LU Hua-xiang (Lab of Neural Networks, Institute of Semiconductor, Chinese Academy of Science, Beijing 100083, China) Abstract: A stochastic model based on conditional probability and Gibbs sampling is proposed to cope with the mod- eling problems occurred in traditional algorithms for distribution estimation, and extends the generality of the algorithm. The algorithm with this model takes promised individuals in the evolution process to form supervised training sets. For each of such sets, we estimate the conditional probability of a component given other components, and execute a Gibbs sampling procedure to generate new candidates for replacing inferior ones. The result of computer experiments shows that the improved algorithm can obtain the global optimum of additively decomposed functions, demonstrating a strong ability in global optimization. Key words: estimation of distribution algorithm; Gibbs sampling; classification; supervised learning 1 (Introduction) . , (evolutionary computation, , evolutionary algorithm) , , (estimation of distribution al- , , gorithms, EDAs)[1–2], , , [3–6]