Course catalogue doctoral education - VT24

    Startpage
  • Ansökan kan ske mellan 2023-10-16 och 2023-11-15
Application closed
Print
Title Introduction to Machine Learning
Course number 5650
Programme 0-Inte del av forskarutbildningsprogram
Language English
Credits 1.5
Date 2023-04-24 -- 2023-04-28
Responsible KI department Institutet för miljömedicin
Specific entry requirements Epidemiology I and Biostatistics I or corresponding courses where basic statistical concepts and linear regression are introduced.
Purpose of the course The purpose of this course is to give an introduction to machine learning without heavy-mathematics. A main focus is on machine learning algorithms for regression analyses using large datasets, both in terms of the number of variables observed and/or the number of units (sample size).
Intended learning outcomes After successfully completing this course, the student is expected to be able to:

- Recognize and formulate well defined questions that can be solved using machine learning algorithms: both prediction and inference questions.
- Explain key concepts in machine learning, including curse of dimensionality, out of sample validation, generalization, uncertainty.
- Choose relevant machine learning algorithms for prediction and inference.
- Conduct and interpret simple analyses using machine learning algorithms on real data.
Contents of the course This course focuses on machine learning algorithms for regression analyses using large datasets, both in terms of the number of variables observed and/or the number of units (sample size). Register data studies are a typical example where such large datasets are analysed. The course will start by going through key concepts of statistical learning, including regression and prediction problems, curse of dimensionality, out of sample validation, generalization, uncertainty. These are problems and concepts that students need to be able to recognize, formulate and explain. The course will then present some central machine learning algorithms, including Lasso, trees, random forest, bagging, together with methods to validate the algorithms and to draw inference. The students will in particular learn how to choose relevant algorithms for specific situations.
Teaching and learning activities Lectures, quizzes, group sessions and computer labs.
Compulsory elements Individual examination (summative assessment).
Examination To pass the course, the student must show that the learning outcomes have been achieved. Assessments methods used are group tasks (formative assessments) along with a written individual task (summative assessment). The examination is viewed as a contribution to the development of knowledge, rather than as a test of knowledge. Students who do not obtain a passing grade in the first examination will be offered a second examination within two months of the final day of the course. Students who do not obtain a passing grade at the first two examinations will be given top priority for admission the next time the course is offered.
Literature and other teaching material Suggested reading:

An introduction to statistical learning by James, Witten, Hastie and Tibshirani. https://www.statlearning.com
Number of students 8 - 25
Selection of students Eligible doctoral students are prioritized according to 1) the relevance of the course syllabus for the applicant's doctoral project (according to written motivation), 2) date for registration as a doctoral student. Give all information requested, including a short description of current research training and motivation for attending, as well as an account of previous courses taken. Prior knowledge in any software is recommended.
More information Please bring your own laptops to class.
Additional course leader The course is given as an elective course within the Swedish Interdisciplinary Graduate School in register-based research (SINGS), website: https://ki.se/en/imm/sings. The course will be given at Umeå University, Umeå. Course leader is Xavier de Luna, PhD, Professor of Statistics, Umeå School of Business, Economics and Statistics, Unit of Statistics, Umeå.
Latest course evaluation Not available
Course responsible Anita Berglund
Institutet för miljömedicin

Anita.Berglund@ki.se
Contact person Johanna Bergman
Institutet för miljömedicin

johanna.bergman@ki.se

Nobels väg 13

17177
Stockholm