In this chapter, we've seen how to apply unsupervised learning algorithms to neural networks. We've been presented a new and suitable architecture to that end, the SOMs of Kohonen. Further, unsupervised learning has been proven to be as powerful as the supervised learning methods because it concentrates only on the input data, without the necessity of input–output mappings. We've seen two new training algorithms: competitive learning and its extension for a Kohonen network. The SOMs also play a role in clustering and dimensionality reduction, besides providing a graphical representation of large datasets. With the content learned so far, we can move to the next chapter , which discusses an interesting practical application of weather forecasting.