三种新型神经网络树算法之改进与比较分析

VittorioMurino提出了三种将神经网络和树结构结合起来的算法。简单介绍了三种结构算法并对其做了细化改进，使其实现更具体、可行，应用范围更广泛；分别利用两组模拟数据和Iris植物分类数据对三种结构与传统的神经网络做了比较实验，给出了三种结构用于模式识别的算法速度、识别率、泛化能力等性能的对比分析。结果表明，三种各具特色的神经网络树结构均有显著的优越性。

Modification and Comparison Analysis of the Algorithms Based on Three Novel Neural Network Structures

Vittorio Murino puts forward three algorithms based on the combination of neural network and tree structure.In this paper,the three algorithms are simply introduced,then specified and modified so as to make its implementation more concrete,feasible and also its application scope more extensive.Two groups of simulative data and Iris botany classification data are respectively used to make experiments to compare these three structures with the conventional MLP neural network,the performances of the three structures in application of pattern recognition such as algorithm velocity,recognition rate,genelization ability and so on are analyzed for comparision.The results show that these three individual algorithms have significant advantages.