8 Fundamental unsupervised learning algorithms
In previous chapters, we looked at supervised algorithms for classification and regression; in this chapter, we shift our focus to unsupervised learning algorithms. Unsupervised learning takes place when no training labels are available. In this case, we are interested in discovering patterns in data and learning data representations. Applications of unsupervised learning span from clustering customer segments in e-commerce to extracting features from image data. In this chapter, we’ll start by looking at the Bayesian nonparametric extension of the K-means algorithm followed by the EM algorithm for Gaussian mixture models (GMMs). We will then look at two different dimensionality reduction techniques—namely, PCA and t-SNE—used to learn an image manifold. The algorithms in this chapter were selected for their mathematical depth and usefulness in real-world applications.