About this course

DAT300 builds upon subjects students have learned in DAT200 - Applied Machine Learning. The covered methodology may include:

  • Foundations of artificial neural networks (NN)
  • Deep convolutional neural networks (CNN)
  • Recurrent neural networks (RNN)
  • Autoencoders
  • Generative models (Variational Autoencoders, Introduction to Generative Adversarial Networks, and Transformers).

The course will introduce the theoretical basics of the methods discussed and focus heavily on applications and modelling with real data. The students will learn to build effective and accurate models that, depending on the application, can contribute to several of UN's sustainability goals, among others 3, 11, 12, 14, 15.

Learning outcome

Good skills and insight into basic techniques for machine learning and deep learning. Basic understanding of various models' mathematical properties and operations. The student will learn to master analysis methods suited for 1) General purpose Machine Learning, 2) Sequence analysis, 3) image classification. The student will learn to connect problem characteristics with choice of appropriate learning algorithms.

  • Learning activities

    The course will consist of lectures and practical exercises using computers and modern machine learning / deep learning software (with help from teaching assistants).

  • Teaching support

    Machine learning / deep learning is a subject that constantly evolves, and online learning resources will be connected to lectures and exercises through the course webpages in Canvas.

    The students can also request appointments with the lecturer in his/her office on pre-arranged times and via email.

  • Syllabus

    Curriculum, programs, supporting literature, etc. will be announced on the course web page.

  • Prerequisites

    DAT200 or similar.

    INF201/200 or a similar course in advanced programming.

  • Recommended prerequisites

    MATH123/280

  • Assessment method

    Written exam, 3.5 hours. A-F.

    Written exam Karakterregel: Letter grades Hjelpemiddel: A1 No calculator, no other aids

  • About use of AI

    K1- No use of AI.

    The AI tools are not allowed to be used during the course assignment, compulsory submissions, and during the examination.

    Descriptions of AI-category codes.

  • Examiner scheme

    An external censor will participate together with the internal censor in forming the exam and censor guide. The external censor checks the internal censor's assessment of a random selection of candidates as a calibration at certain intervals in line with the faculty's guidelines for censoring.

  • Mandatory activity

    Compulsory hand-in assignments. Rules for approving obligatory activities will be announced when the course starts.

  • Teaching hours

    Lectures: 4 hours per week. Exercises: 2 hours per week.