In this section, we present unsupervised learning models in neural network, named competitive learning and Kohonen SOM. Kohonen SOM was invented by a professor named Teuvo Kohonen and is a way to represent multidimensional data in much lower dimensions: 1D or 2D. It can classify data without supervision. Unsupervised learning aims at finding hidden patterns within the dataset and clustering them into different classes of data.
There are many unsupervised learning techniques, namely K-means clustering, dimensionality reduction, EM, and so on. The common feature is that there is no input-output mapping and we work only on the input values to create a group or set of outputs.
For the case of neural networks, they can be used for unsupervised learning. They can group data into different buckets (clustering) or abstract original data into a different set of output data points (feature abstraction or dimensionality reduction). Unsupervised techniques require...