An Improved Multi-SOM Algorithm

Provided by: University of Tulsa
Topic: Big Data
Format: PDF
Clustering is an unsupervised learning technique aiming to obtain homogeneous partitions of objects while promoting the heterogeneity between partitions. In the literature there are many clustering categories such as hierarchical, partition-based, and density-based and Neuronal Networks (NN). Hierarchical methods aim to build a hierarchy of clusters with many levels. There are two types of hierarchical clustering approaches: the agglomerative methods (bottom-up) and the divisive methods (Top-down).Agglomerative methods start by many data objects taken as clusters and are successively joined two by two until obtaining a single partition containing all the objects.

Find By Topic