Practical Machine Learning Cookbook

Chapter 7. Unsupervised Learning

In this chapter, we will cover the following recipes:

• Self-organizing map - visualizing heatmaps
• Vector quantization--Image clustering

Introduction

Self-organizing map (SOM): The self-organizing map belongs to a class of unsupervised learning that is based on competitive learning, in which output neurons compete amongst themselves to be activated, with the result that only one is activated at any one time. This activated neuron is called the winning neuron. Such competition can be induced/implemented by having lateral inhibition connections (negative feedback paths) between the neurons, resulting in the neurons organizing themselves. SOM can be imagined as a sheet-like neural network, with nodes arranged as regular, ...