| 基于改进遗传算法的神经网络优化 |
| |
| 引用本文： | 镇方雄,李跃新.基于改进遗传算法的神经网络优化[J].湖北大学学报(自然科学版),2006,28(4):345-349. |
| |
| 作者姓名： | 镇方雄 李跃新 |
| |
| 作者单位： | 咸宁学院数学系 湖北咸宁437000(镇方雄)，湖北大学数学与计算机科学学院 湖北武汉430062(李跃新) |
| |
| 基金项目： | 塔里木油田乙方队伍考核网资助项目 |
| |
| 摘 要： | 针对在神经网络应用中,存在结构设计及权值训练算法的不足,提出一种新的基于混合编码方案的遗传算法.在算法中设计了用遗传算法全局优化神经网络拓扑结构和网络权值的新的编码方案,改进了适应度函数的设计和采用自适应的交叉和变异方法.试验结果表明本算法能有效地对神经网络的权值和结构同时优化,提高了训练效率.
|
| 关 键 词： | 神经网络 遗传算法 拓扑结构 权值 |
| 文章编号： | 1000-2375(2006)04-0345-05 |
| 收稿时间： | 06 29 2005 12:00AM |
| 修稿时间： | 2005年6月29日 |
Neural network optimize based on improved genetic algorithm |
| |
| ZHEN Fang-xiong,LI Yue-xin.Neural network optimize based on improved genetic algorithm[J].Journal of Hubei University(Natural Science Edition),2006,28(4):345-349. |
| |
| Authors: | ZHEN Fang-xiong LI Yue-xin |
| |
| Institution: | 1. Department of Mathematics, Xianning College, Xianning 437000, China; 2. School of Mathmatics and Computer Science, Hubei University, Wuhan 430062, China |
| |
| Abstract: | In the application of neural network,it exists the shortage of the structure design and weitht training algorithm,a new genetic algorithm based on the mixed encoding scheme is proposed in this paper.In this algorithm,the new coding scheme of the global optimization of the topology and weight distribution of neural network fusing with genetic algorithm is designed,fitness function is improved,adaptive crossover and mutation are used.The result of experiment indicates the algorithm can efficiently optimize the topology and weight of neural network. |
| |
| Keywords: | neural network genetic algorithm structure weight |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
