| 多子群粒子群集成神经网络 |
| |
| 引用本文： | 潘希姣. 多子群粒子群集成神经网络[J]. 安徽建筑工业学院学报, 2007, 15(2): 38-40. DOI: 10.3969/j.issn.1006-4540.2007.02.012 |
| |
| 作者姓名： | 潘希姣 |
| |
| 作者单位： | 安徽交通职业技术学院汽车与机械工程系,合肥,230051 |
| |
| 摘 要： | 为了使参加神经网络集成的个体差异度较大,从而提高网络集成的泛化能力,本文提出一种新的基于多子群粒子群算法的神经网络集成方法.每个子群通过补充差异度独立训练出一批神经网络,从每个子群中选择一个最优个体参加网络集成,实验使用了UCI标准数据集.实验证明,该算法的识别能力要好于Boosting、Bagging等传统方法.
|
| 关 键 词： | 神经网络集成 粒子群优化算法 |
| 文章编号： | 24448066 |
| 修稿时间： | 2007-01-23 |
Artificial neural network ensemble based on multi-sub-swarm particle swarm optimization |
| |
| PAN Xi-jiao. Artificial neural network ensemble based on multi-sub-swarm particle swarm optimization[J]. Journal of Anhui Institute of Architecture(Natural Science), 2007, 15(2): 38-40. DOI: 10.3969/j.issn.1006-4540.2007.02.012 |
| |
| Authors: | PAN Xi-jiao |
| |
| Affiliation: | Dept. of Automobile and Machinery Engineering, Anhui College of Vocational Technology of Communications, Hefei 230051, China |
| |
| Abstract: | To improve the generalization capability of neural network ensemble and enlarge the diversity of every individual artificial neural network(ANN),a novel ANN ensembles based on multi-sub-swarm particle swarm optimization were proposed.Each sub-swarm trains different ANN through complementary diversity.The best individual was chosen from each sub-swarm to take part in the ensemble.The experiments used UCI benchmark datasets,the experimental results show that the recognition rate of this algorithm is better than the Boosting and Bagging algorithm. |
| |
| Keywords: | Artificial neural network ensemble Particle swarm optimization |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
