摘要
分布估计算法是一种新的种群进化算法,通过建立概率模型得到新的个体,copula分布估计算法是将copula理论与分布估计算法结合,提高估计的精确性和效率。针对分布估计算法全局收敛的特点,与BP算法结合可以避免BP算法易陷入局部极值点的缺陷,同时可以使优化结果更加精确。本文采用copula EDA与BP算法的两种结合模式来优化神经网络的权值和阈值,并且比较两种结合模式。可以得出,copula分布估计算法与BP算法融合可以提高收敛速度和精确性。
Estimation of distribution algorithm is a new population evolutionary algoriths. A new individual was ob- tained through the establishment of probabilistic modets. Copula estimation of distribution algorithm combines copu- la theory with estimation of distribution algorithms so as to improve the estimation accuracy and efficiency. Estima- tion of distribution algorithms has global convergence characteristics, its combination with BP algorithm can avoid falling into local extreme-point defects, and the optimization results are more precise. Two binding modes of adop- ting BP algorithm with copula EDA are used to optimize the neural network weights and thresholds, and the two binding modes are compared. The resuhs show that copula estimation of distribution algorithm and BP algorithm can improve the speed and accuracy of convergence.
出处
《太原科技大学学报》
2014年第1期28-33,共6页
Journal of Taiyuan University of Science and Technology
基金
太原科技大学博士启动基金(20122009)
太原科技大学研究生创新项目(20125012)
关键词
copula分布估计算法
BP算法
神经网络
优化
Estimation of distribution algorithm based on copula, BP algorithm, neural network, optimization
