摘要
该文针对竞争型神经网络在对训练样本没有明显的分类特征时不能准确分类的问题,在已有的分类算法基础上,提出了一种改进的分类算法。该算法在竞争型神经网络学习过程,通过对训练样本引入特征向量的方法,进一步改善了分类效果。实验结果表明,该算法在分类过程中表现了良好的效果,不仅降低了训练误差,还具有很好的分类准确率,表现出很好的学习效率,通过对比说明该算法的有效性和优越性。
Aiming competitive neural network training samples when there is no obvious feature is not an accurate classification classification problem, in the existing classification algorithm is proposed based on an improved classification algorithm. The competitive neural network algorithm in the learning process, the method of training samples by introducing feature vectors, to further improve the classification results. Experimental results show that the algorithm performance in the classification process with good results, not only reduces the training error, but also has good classification accuracy, showing good learning efficiency by comparing the description of the algorithm effectiveness and superiority.
作者
南书坡
郭战杰
程聪
韩利华
NAN Shu-po1,GUO Zhan-jie2,CHENG Cong1,HAN Lin-hua1（1 .Henan Normal University, College of XinLian, ZhengZhou 451）001）,China; 2.Zhengzhou Vocational and Technical College , Zhengzhou 450000,China）
出处
《电脑知识与技术》
2017年第6期138-140,共3页
Computer Knowledge and Technology
关键词
竞争型神经网络
分类
训练误差
特征向量
competitive neural network
classification
training error
eigenvectors
