Machine Learning

Doel vak

Upon completion of this course, students will:

• be acquainted with the dominant concepts of machine learning methods, including some theoretical background. (Knowledge and understanding),
• acquire knowledge of established machine learning techniques such as linear models, neural networks, decision trees and ensemble methods (Knowledge and understanding),
• be able to use machine learning methods effectively in code, including the ability to machine learning experiments to tune and select models effectively, (Apply knowledge and understanding, make judgments),
• know some statistical techniques to assess and validate machine learning results. (Apply knowledge and understanding, make judgments)

Inhoud vak

Machine learning is the discipline that studies how to build computer programs that learn how to behave from examples, rather than following explicit instructions. It is a subfield of artificial intelligence that intersects with statistics, cognitive science, information theory, and probability theory.

Recently, machine learning has become increasingly important for the design of search engines, robots, and sensor systems, and for the processing of large scientific data sets. Other applications include handwriting or speech recognition, image classification, medical diagnosis, stock market analysis and bioinformatics. Machine Learning is also the core technology behind chatbots like ChatGPT, Claude and Gemini.

The course covers a wide variety of machine learning techniques, but puts particular emphasis on gradient descent optimization, backpropagation, neural networks and deep learning. Some discussion on the broader social impact of machine learning technology is included.

Aanvullende informatie onderwijsvormen

The course consists of pre-recorded videos, interactive lecture/QA sessions (two per week) and optional homework assignments discussed in working groups (one per week). The practical assignment is supported by small exercises to help with the relevant technologies, and informal presentations at project groups (one per week).

There is no mandatory attendance for any lectures or workgroups, except that one member of each group must be present at the weekly project group. Most of the material is freely available at https://mlvu.github.io The course is taught in English.

Toetsvorm

The course assessment consists of two parts: an exam and a practical assignment.

The examination consists of two parts: a mid-term exam including open-ended essay questions and a fully multiple-choice final exam.

The exam and practical assignment each comprise 50% of the final grade.
To pass the course, the exam grade should be at least 5.5, the practical assignment grade should be at least 4.5 and the average should be at least 5.5.

The exam is made individually, and the practical assignment is made in groups. There are separate resits for both parts of the exam, no resit is possible for the practical assignment.

Literatuur

Aanvullende informatie doelgroep

This course is currently part of the curriculum of:
Bachelor Artificial Intelligence
Bachelor Business Analytics
Bachelor Computer Science

Premaster Artificial Intelligence
Premaster Business Analytics

It is available as an elective to students of:
Bachelor Mathematics
Bachelor Medische Natuurwetenschappen

If you are not in these programs, registration is allowed, but be sure that you have mastered the required preliminaries

If you are in the MSc Bioinformatics, you need to register for Scientific Machine Learning (XM_0138), not this course.

Overige informatie

In both the writing and coding for the final report the AI policy is as follows:

You may consult with AI tools during programming and writing, but you may not have the AI system write code or text for you. You are held personally responsible for everything you hand in.

The same philosophy applies to other phases of the process like ideation, experiment design and debugging. In all cases, you may consult an AI chatbot, but you may not defer to the AI chatbot. This policy is explained in detail, with examples, on the course Canvas page.

All AI use must be declared. Undeclared AI use of any kind is considered fraud.

In the case of serious suspicions of AI fraud, an in-person follow-up interview may be required to obtain the grade, to establish that you understand everything you handed in.

Afwijkende intekenprocedure

Workgroup registration will be done through Canvas in the first week of the course. Only a registration for the main course itself is required to take part in all parts of the course.

Aanbevolen voorkennis

We require that students have some prior experience with linear algebra, calculus (limited to differentiation), probability theory and statistics. An overview and explanation of the required preliminaries can be found at https://mlvu.github.io/preliminaries/

A basic understanding will suffice and we will take some time to go over the basics again. Feel free to register if you have no experience with any of these, but expect to put in a little extra effort in the first weeks (using the materials provided).

Programming experience in python is also required.