This completes the overview of some of the techniques to reduce the complexity and dimension of a problem.

We learned that the simplicity of divergences makes these techniques attractive for feature extraction and dimension reduction on a smaller set of features. Principal component analysis is a robust dimension reduction technique for linear or pseudo-linear models. Finally, manifold learning for nonlinear models is a technically challenging field with great potential in terms of dynamic object recognition [5:11].

In the next chapter, we will address supervised learning techniques, starting with generative models.