一种优化神经网络结构的遗传禁忌算法  

Genetic and tabu search algorithm for the optimization of neural network structure

作　　者：王淑玲[1] 李振涛[1] 邢棉[1] 

机构地区：[1]华北电力大学数理系,河北保定071003

出　　处：《计算机应用》2007年第6期1426-1429,共4页journal of Computer Applications

基　　金：国家自然科学基金资助项目(50077007)

摘　　要：常用的神经网络是通过固定的网络结构得到最优权值,使网络的实用性受到影响。引入一种基于方向的交叉算子和禁忌变异算子,同时把禁忌算法(TS)引入标准遗传算法,结合标准遗传算法和禁忌算法的优点,提出一种优化神经网络结构的遗传禁忌混合算法,实现了网络结构和权值同时优化。仿真实验表明,与遗传算法和禁忌算法相比,该算法优化的神经网络收敛速度较快、预测精度较高,提高了网络的处理能力。A conventional Neural Network often optimizes the weights through invariable network structure, which has limited the extensive use of the Neural Network. The crossover operator based on direction and Tabu search mutation operator was introduced. This paper put forward Genetic and Tabu search algorithm to train the neural networks, combining the merits of genetic algorithm and that of Tabu search algorithm, which makes weights and structure of artificial neural networks be optimized together. The result shows that the neural network optimized by using the presented algorithm has the advantages of quicker convergence rate and higher precision, compared with genetic algorithm and Tabu search algorithm, and that the processing ability of networks is also raised.

关 键 词：遗传禁忌算法 神经网络 优化 算子 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]