This is the homepage of SOM Toolbox, a function package for Matlab 5 implementing the Self-Organizing Map (SOM) algorithm and more. You can

• train SOM with different network topologies and learning paramteres,
• compute different error, quality and measures for the SOM,
• visualize SOM using u-matrices, component planes, cluster color coding and color linking between the SOM and other visualization methods, and
• do correlation and cluster analysis with SOM.

SOM Toolbox also features other data analysis methods related to VQ, clustering, dimension reduction, and proximity preserving projections, e.g.,

• data preprocessing tools,
• K-means, K-nearest neighbor classifier and LVQ (learning vector quantizer),
• agglomerative hierarchical clustering and dendrograms,
• principal component analysis (PCA),
• Sammon's projection, and
• Curvilinear Component Analysis (CCA).

---

A short introduction to SOM by Teuvo Kohonen.

---

SOM Toolbox 2.0

• About SOM Toolbox
• Download (see news below for version 2.1)
• Documentation
• Gallery

Related

• SOM_PAK and LVQ_PAK homepage at CIS.
• Related publications, projects, software, demos,...

Latest news:

Jan, 2015

SOM Toolbox v2.1 is available in github and open for contributions. See this page for details.

Mar 23th, 2005

You will probably run into problems with Matlab 7...>>>

Mar 17th, 2005

Bug corrections>>> (e.g., som_fuzzycolor), a brush up of the WWW pages.

August 7th, 2003

Some modifications have been done to be compatible with Matlab 6.5.

---

Kindly observe that the members of the SOM Toolbox team have left CIS or are involved with other duties. Therefore, we cannot currently guarantee any support in using SOM Toolbox.