| 演化神经网络算法 |
| |
| 引用本文： | 蔡晓芬,方建斌.演化神经网络算法[J].江汉大学学报(自然科学版),2005,33(3):18-21. |
| |
| 作者姓名： | 蔡晓芬 方建斌 |
| |
| 作者单位： | 1. 上海工程技术大学,基础教学学院,上海,200336
2. 江汉大学数学与计算机科学学院,湖北,武汉,430056 |
| |
| 摘 要： | 演化算法的应用使神经网络的训练有了一个崭新的面貌,目标函数既不要求连续,也不要求可微,仅要求该问题可计算,而且它的搜索始终遍及整个解空间,因此容易得到全局最优解.应用这一思想,本文提出了一种新的演化神经网络算法—EPANN,并通过算法实例说明EPANN是可行的.
|
| 关 键 词： | 演化算法 遗传规划 人工神经网络 结构变异 最优解 |
| 文章编号： | 1673-0143（2005）03-0018-04 |
| 收稿时间： | 2005-02-30 |
| 修稿时间： | 2005年2月28日 |
Evolutionary Neural Network Algorithm |
| |
| CAI Xiao-fen,FANG Jian-bin.Evolutionary Neural Network Algorithm[J].Journal of Jianghan University:Natural Sciences,2005,33(3):18-21. |
| |
| Authors: | CAI Xiao-fen FANG Jian-bin |
| |
| Abstract: | The application of evolutionary algorithm gives the training of neural network a new feature,because it only needs the question is calculable and without the consecution and the differ- ential calculus of the objective functions, its ’very easy to get the superior solution of the overall situ- ation because of its overall situation searching. A new evolutionary neural network algorithm— EPANN is presented , and the experimental results show that EPANN has great computation ability in comparison with other algorithms. |
| |
| Keywords: | evolutionary algorithm genetic programming artificial neural network structure variation superior solution |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
