论文部分内容阅读
本文提出了一种基于自组织特征映射的前债式人工神经网络模型，介绍了其结构和算法．该模型基于自组织特征映射机理，用统计方法获得输入信号对不同模式类别的隶属程度，并由此进行模式分类判决计算．该神经网络模型还导出了“模式地形图”的概念，可以实现数据聚类分析的可视化．经计算机模拟验证，上述算法和概念是有效的．
In this paper, a model of pre-debt artificial neural network based on self-organizing feature map is presented, and its structure and algorithm are introduced. Based on the self-organizing feature mapping mechanism, the model obtains the degree of membership of input signals to different pattern categories by using statistical methods and calculates the pattern classification decisions. The neural network model also leads to the concept of “pattern topographic map”, which can visualize data clustering analysis. The above algorithms and concepts are validated by computer simulation.