In this chapter, we introduce principal component analysis (PCA), a common practice to reduce its dimensionality, and discuss the link between PCA and low-rank approximations.

This chatper first introduces the kernel trick, which allows us to operate in the original lower-dimensional domain. We then discuss decision tree and ensemble methods for reducing data over-fitting.

This chapter introduces the numerical convex optimization problem that minimize a certain objective function subject to some constraints. We also introduce an efficient algorithm for solving such problems.

This chapter introduces the classical mean/variance portfolio design approach, and discusses extensions of the basic model, including transaction costs, market impact, and risk beyond the variance.

This chapter provides an overview of the topics covered in this, the book’s structure, the scope and presentation of the books, and the target audience for the book.

In this chapter we introduce the main concepts of neural networks (NNs). Next, we present the main building blocks of a neural network and we discuss the most common training techniques.

In this chapter, we focus on linear classifiers such as logistic regression, and linear support vector machines. We then extend to the multi-class case and discuss issues of regularization, sparsity, robustness, and class imbalanace.

When speaking, speakers break down the continuous stream of sounds into smaller units – so-called intonation phrases. Within these intonation phrases, we find a rhythmic pattern of unstressed and stressed syllables, similar to the word stress that we already saw in individual words. This chapter investigates which principles speakers follow when rhythmically grouping the speech flow. Like word stress, the rhythmic structure of utterances does not distinguish meaning in English, or it only does so in a very limited way. However, it does serve important linguistic functions. Additionally, speakers pronounce intonation phrases with specific melodic patterns. Intonation will be dealt with in detail in the subsequent chapter. Rhythm and intonation constitute the suprasegmental level. Even though the suprasegmental level plays a secondary role in the language system of English, rhythm and intonation serve several crucial functions in communicative situations.

In this chapter, we are going to ask one of the most important questions that phonology is concerned with: when are two phonetically different sounds considered variations of the same sound within the system of a language and when are they two distinct speech sounds? Instead of speech sounds, we are going to introduce the terms ‘phonemes’ and ‘allophones’ and use these from hereon. We will learn how phonemes and allophones differ in their function, their occurrence within language (called ‘distribution‘) as well as their phonetic properties. Finally, we can now understand the different interests that the two sister-sciences phonetics and phonology pursue.

Properties of speech sounds can be measured. Speech recognition systems, such as Alexa, Siri and Google Assistant, make use of the fact that each class of sounds has its own characteristics. In this chapter, we are going to look at the most basic acoustic characteristics of speech, starting with loudness and pitch. Moreover, you will learn how you can make easy measurements with a free software that many phoneticians also use.

In this chapter, you will discover that sound-changing processes are also at work in connected speech that impact the phoneme sequence. Like allophonic variation, many of these will occur in faster speech and not in careful pronunciation, and they are also speaker-dependent. Over time, though, they may alter the pronunciation of a word, so that many of today’s pronunciations are the result of such phonological processes that occurred in the past, and current frequently occurring phonological processes will lead to the standard pronunciation of tomorrow.

In this chapter, we will look at the writing system, why the Latin alphabet used by English and many other languages is not ideal, as it does not provide letters for all sounds, and what tools phoneticians have at their disposal to accurately write down the pronunciation of words. This textbook takes a hands-on approach. Therefore, in this chapter, you will also be introduced to a free software used by many phoneticians. With this, you can listen to stretches of speech and annotate them. In a later chapter, you will also learn how to perform simple measurements with it. Finally, just like there is standard English grammar and vocabulary, there are also two main pronunciation standards for English, which will be introduced here.

This chapter introduces diphthongs as well as other English vowels that only occur in specific positions. Unlike for monophthongs, where the tongue remains in a steady position, for diphthongs the tongue performs a gliding movement during articulation. Diphthongs are an integral part of the English vowel system. Sometimes diphthongs are indicated in the spelling by a digraph, such as in they. At other times, however, they are represented by a single vowel letter, as in gave. Two classes of vowels deserve special attention, as they only occur in specific positions. Pre-rhotic vowels only occur when in spelling the vowel is followed by ‹r›. Weak vowels, on the other hand, only appear in unstressed syllables, so in syllables that are not accentuated. Finally, the chapter looks at letter-to-sound correspondences. These are more intricate, as the thirteen vowel qualities found in English can only be written with the five vowel letters of the Latin alphabet. However, they are also more regular than might be assumed.