There is no silver bullet: no model can fit all data. Hence, special data requires special algorithms. In this chapter, we deal with two types of special data: sparse data and sequences that can be aligned to each other. We will not dive deep into sparsity learning, which is very complex. Rather, we introduce key concepts: sparsity inducing loss functions, dictionary learning, and what exactly the word sparsity means. For the second part in this chapter, we introduce dynamic time warping (DTW), which deals with sequences that can be aligned with each other (but there are sequences that cannot be aligned, which we will discuss in the next chapter). We use our old tricks: ideas, visualizations, formalizations, to reach the DTW solution. The key idea behind its success is divide-and-conquer and the key technology is dynamic programming.

The normal distribution is the most widely used continuous distribution, but many of its relevant properties are a little bit advanced for an undergraduate course. Hence, Part IV introduces some of these advanced topics. This chapter devotes itself to properties of normal distributions: single- and multivariate normal distributions, moment and canonical parameterizations, sum and product, geometry and the Mahalanobis distance, and conditional distributions. We also show that with these properties, some algorithms will become much easier to understand. We use parameter estimation and the Kalman filter as two such examples.

We cannot miss deep learning in a modern pattern recognition textbook, and we introduce CNN (convolutional neural networks) in this chapter. Although the mathematical derivation of CNN, especially the back-propagation process and gradient computation, is complex, we use a lot of useful tools to help readers understand what exactlyis going on in a CNN. Hence, this chapter focuses on accessibility rather than completeness. In its exercise problems, we introduce more relevant topics and methods.

Unlike PCA, which is unsupervised, FLD uses labels associated with data points, and no doubt it may get better linear features and accuracy than PCA. We start by illustrating this motivation, and practice the problem-solving framework by gradually developing the correct mathematical formulation behind the relatively simple idea behind Fisher's linear discriminant (FLD). We discuss various practical issues: the solution for the binary case, the scenario where this solution breaks down, and how to generalize from tasks with only two categories to many categories.

This chapter is a succinct introduction to basic probabilistic methods for pattern recognition and machine learning. One focus is to clearly present the exact meanings of different terms, including the taxonomy of different probabilistic methods. We present a basic introduction to maximum likelihood and maximum a posteriori estimation, and a very brief example to showcase the concept of Bayesian estimation. For the nonparametric world, we start from the drawbacks of parametric methods, gradually analyzing the properties preferred for a nonparametric one, and finally reach the kernel density estimation, a typical nonparametric method.

This chapter is an overall introduction to the definition of pattern recognition, its relationship with machine learning and other relevant subject areas, and the main components and development process inside a pattern recognition system. This introduction is started by considering an autonomous driving example.

Parameter estimation is generally difficult, requiring advanced methods such as the expectation-maximization (EM). This chapter focuses on the ideas behind EM, rather than its complex mathematical properties or proofs. We use the Gaussian mixture model (GMM) as an illustrative example to find what leads us to the EM algorithms, e.g., complete and incomplete data likelihood, concave and nonconcave loss functions, and observed and hidden variables. We then derive the EM algorithm in general and its application to GMM.

This chapter presents a simple but working face recognition system, which is based on the nearest neighbor search algorithm. Albeit simple, it is a complete pattern recognition pipeline. We can then examine every component in it, and analyze potential difficulties and pitfalls one may encounter. Furthermore, we introduce a problem-solving framework, which will be useful in the rest of this book and in solving other tasks.

This chapter is not about one particular method (or a family of methods). Instead, it provides a set of tools useful for better pattern recognition, especially for real-world applications. They include the definition of distance metrics, vector norms, a brief introduction to the idea of distance metric learning, and power mean kernels (which is a family of useful metrics). We also establish by examples that proper normalizations of our data are essential, and introduce a few data normalization and transformation methods.

Starting from this chapter, Part III introduces several commonly used algorithms in pattern recognition and machine learning. Support vector machines (SVM) starts from a simple and beautiful idea: large margin. We first show that in order to find such an idea, we may need to simplify our problem setup by assuming a linearly separable binary one. Then we visualize and calculate the margin to reach the SVM formulation, which is complex and difficult to optimize. We practice the simplification procedure again until the formulation becomes viable, briefly mention the primal--dual relationship, but do not go into details of its optimization. We show that the simplification assumptions (linear, separable, and binary) can be relaxed such that SVM will solve more difficult tasks---and the key ideas here are also useful in other tasks: slack variables and kernel methods.