matlab神经网络4：数据聚类-自组织映射

最新推荐文章于 2024-07-04 13:22:34 发布

沈子恒

最新推荐文章于 2024-07-04 13:22:34 发布

阅读量6.6k

点赞数 1

分类专栏：神经网络|深度学习|matlab

本文链接：https://blog.csdn.net/shenziheng1/article/details/79174174

版权

该博客探讨了使用MATLAB进行数据聚类的问题，重点介绍了自组织映射(SOM)算法。首先解释了聚类的基本概念，然后详细阐述了SOM在聚类中的应用，以及如何通过SOM实现数据的无监督学习。此外，还讨论了聚类后的性能评价指标，以评估和优化聚类效果。

摘要由CSDN通过智能技术生成

1.聚类问题与自组织映射

In clustering problems, you want a neural network to group data by similarity. For example: market segmentation done by grouping people according to their buying patterns; data mining（数据挖掘） can be done by partitioning data into related subsets; or bioinformatic analysis such as grouping genes with related expression patterns.

A self-organizing map (selforgmap) consists of a competitive layer which can classify a dataset of vectors with any number of dimensions into as many classes as the layer has neurons. The neurons are arranged in a 2D topology, which allows the layer to form a representation of the distribution and a two-dimensional ap