基于MATLAB的BP-LVQ神经网络组合分类模型

综合考虑神经网络分类误差率以及训练速率,文中从组合分类器结构出发,提出一种树形多层的BP-LVQ神经网络组合分类器模型。该组合分类器利用BP神经网络独立性以及自适应性解决了一般分类器难以不断学习和适应新攻击的问题,利用LVQ神经网络的竞争性将客观分类信息转变成使用者所定义的类别。利用MATLAB神经网络工具箱对该BP-LVQ神经网络组合分类模型进行仿真实验,实验结果表明,该组合分类器的分类效率明显高于单一分类器。

BP-LVQ Neural Networks Classifiers Combination Model Based on MATLAB

By considering the error rates and the training speed of neural networks,a hierarchical classifiers which is called as BP-LVQ neural network combination model is proposed in this paper.Combining with the capability of independency and self-adaptation of BP neural network,this combined classifier can solve the problem that normal classifiers can't keep learning and adapting new attack,and the problem that how to change the impersonal class into the category user defined by competitive position of LVQ neural network.The experimental results show that the tree-schema model can keep better performance than single classifier.