NeuralKits神经网络软件工具箱的开发和验证  

Development and Verification of the NeuralKits Neural Network Software Kit

作　　者：陈鹏 Chen Peng(Fujian Provincial Industrial and Information Industry Development Research Center,Fuzhou,Fujian 350003,China)

机构地区：[1]福建省工业信息产业发展研究中心,福建福州350003

出　　处：《计算机时代》2025年第3期40-44,共5页Computer Era

摘　　要：利用Visual Basic 6.0语言成功开发出NeuralKits神经网络软件工具箱,包括基于ART2自适应谐振理论Ⅱ神经网络、BP误差反向传播神经网络、CPN对偶传播神经网络、PNN概率神经网络、RBF径向基函数神经网络和SOM自组织映射神经网络的相应软件。首先通过工具箱在处理平衡数据(鸢尾花品属数据集)时所展现出的正确率,对其开发和运行的正确性予以验证;再利用R语言编程实现随机森林算法,并对比该算法与工具箱在处理非平衡数据(白酒品质数据集)时各自的正确率,以此间接验证工具箱的正确性。This paper presents the successful development of the NeuralKits neural network software kit using Visual Basic 6.0.The kit contains ART2 neural network,BP neural network,CPN neural network,PNN neural network,RBF neural network and SOM neural network.Correctness of development and performances of the kit was first verified by processing balanced data(the iris dataset)to obtain relevant accuracy and finally indicated by using the random forest algorithm programmed in R language and comparing it with the kit when dealing with unbalanced data(the white wine quality dataset)to compare the accuracies.

关 键 词：NeuralKits Visual Basic 6.0 ART2神经网络 BP神经网络 CPN神经网络 PNN神经网络 RBF神经网络 SOM神经网络 随机森林 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论]