基于多维自组织特征映射的聚类算法研究

Study of Algorithms of Clustering Based on Multi-dimensional Self-organizing Feature Mapping

作为神经网络的一种方法,自组织特征映射在数据挖掘、模式分类和机器学习中得到了广泛应用。本文详细讨论了自组织特征映射的聚类算法的工作原理和具体实现算法。通过系统仿真实验分析,SOFMF算法很好地克服了许多聚类算法存在的问题,在时间复杂度上具有良好的性能。

As a method of neural network, the self-organizing feature mapping（SOFM） is an excellent approach for data mining, pattern classification and machine learning. The theory and algorithm of SOFM are discussed in detail in this article. Simultaneously analyze and summarize this algorithm, overcome the insufficiency of many clustering algorithms, be able to find clusters in different shapes, be non-sensitive to the input data sequence, process noise data and multi-dimensional data well, and have multi-resolution.