Welcome to this course on the Foundations of Machine Learning! This is a self-study course based on the text book Machine Learning: A first course for engineers and scientists (Lindholm et al., 2021). The book is freely available in pdf format by following this link (by the way, an underlined sentence or word indicates a link). Taking the course involves reading selected chapters from the book and then carrying out exercises on this web-based course platform.

The online material consists of eight sections, each comprising a number of subsections, or exercises. (To avoid confusion we will refer to parts of the course book as chapters, e.g. chapter 3.1, whereas the word section is used to refer to contents on the web.)

This first section will give an introduction to the subject of machine learning, as well as introduce the course platform itself. But, before moving on, feel free to watch the following introductory lecture on machine learning, providing futher information about the course:

Good luck!

Lindholm et al. (2021) describe machine learning as being about “learning, reasoning, and acting based on data”. As will become evident in this course, the subject has strong links to both statistics and computer science. On the one hand, we will use statistical reasoning and a mathematical language to describe the machine learning models that we will encounter. On the other hand, we will use programming and numerical software tools to turn these mathematical descriptions into practical machine learning models.

In practice, a machine learning model is nothing but a computer program that takes some input data and produces an output, which could for instance be a prediction of some unknown quantity of interest or a suggested action to take. As an example, borrowed from the work by Esteva et al. (2017), the input data could correspond to an image of a skin lesion and the objective of the computer program is to determine if the lesion is benign or malignant. Hence, the output of the program is a prediction of what type of lesion that the program is presented with.

What is special about machine learning is that the computer program is not written “from scratch” by some domain expert. Instead, generic programs (or models) are used that can be adapted to data. For instance, in the skin lesion example, the program is trained based on skin lesion images annotated by dermatologists, in order to learn from their expertise and to mimic their diagnostic predictions. Hence, we can view machine learning as a way programming by examples. The beauty of machine learning is that this is done in a generic way, in the sense that the models and learning algorithms can be used in many different contexts and applications domains, ranging from medicine to engineering and social sciences. Chapter 1 of the course book gives additional examples of machine learning applications. It also defines some of the key concepts used to describe and differentiate between different types of machine learning.

Start by reading chapter 1 in the course book. When you are done, you can check the box below and press Submit – this is your first “exercise”. You will see that submitting your solutions to the exercises on the course platform requires that you log in with your LiU ID (see the Log in button in the top right corner of the page, if you are not already logged in). This in turn requires that you are a students at Linköping University. Please see the Course info page for information about how to apply and register for the course.

I've read chapter 1 in the course book.

PS. It is possible to take part of the web-based material without signing in—simply click the Next exercise button below to continue. However, in that case you will not be able to verify your solutions, run code, or store your progress.