摘要
以认知心理学、模型集成理论为基础,构建了集粗糙集理论、聚类理论、模糊逻辑理论、遗传算法理论、人工神经网络理论于一体的一个新的混合智能系统R-FC-DENN。它首先通过粗糙集将输入样本进行约简,然后用聚类技术将简化后的样本进行聚类,对不同的聚类使用经过遗传算法改进了的神经网络进行训练,接着将这些经过不同神经网络训练的样本用模糊权值组合起来,放入新的用遗传算法改进了的神经网络再进行训练,从而完成整个训练过程。最后用UCI下的实际数据库对提出的混合智能系统R-FC-DENN的实用性进行了检验,证明方法是可行的。
Based on the cognitive psychology and aggregative model theory, a new hybrid intelligent system-R-FC-DENN incorporating rough set, clustering theory, fuzzy logic, genetic algorithm and artificial neural network is propased. Firstly, R-FC-DENN uses the rough set to reduce the data, then it clusters the data by the clustering theory. After that, it is the clustered data train with improved ANN. Subsequently, the trained data are fabricated by fuzzy weight. Lastly, the fabricated data are trained by another improved ANN and thus the whole process of training is completed. In the end the UCI＇s databases are employed to prove the utility of the new HIS-R-FC-DENN and a satisfactory result is obtained.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2006年第3期448-453,共6页
Systems Engineering and Electronics
基金
国家自然科学基金资助课题(70571016
70471011)
关键词
混合智能系统
人工神经网络
粗糙集
遗传算法
hybrid intelligent system
artificial neural network
rough set
genetic algorithm
