基于粗糙集的并行协作模块化神经网络模式分类器

A Parallel Cooperative Modular Neural Network Classifier Based on Rough Set

该文提出了基于粗糙集的K类模式分类器的体系结构(RSPCMNNC),基于粗糙集理论提出了三个预处理算法,简化了分类器的结构,降低了学习难度,有效地避免产生过多的子网。样本空间基于最大均衡的策略来划分,保证BP算法在学习过程中的有效性。实验结果表明,该文提出的RSPCMNNC分类器显示出更高的识别率,对于实际应用中多特征模式的识别问题,具有很大的实用价值。

This paper proposes a Parallel Cooperative Modular Neural Network pattern Classifier based on Rough Set(RSPCMNNC)and three preprocessing learning algorithms ,which lead to the simplification of a classifier,reduce the learning complexity,avoid producing too many sub-nets,and guarantee the learning validity of the BP algorithm by the sample space partition and the basis of well-balanced strategy.Experimental results show that the above -mentioned RSPCMNNC has a higher recognition rate than any other method,and it is of great value to the recognition of patterns with multiple characters in practical applications.