Topic outline

  • Teachers:
    Michael Riis Andersen (Sessions 1-4), Markus Heinonen (Session 5), and Arno Solin (Session 6).

    Overview:
    Gaussian processes are a powerful tool for Bayesian nonparametric modelling. This course will give an introduction to the field of Gaussian processes and provide a theoretical background for Gaussian processes including both modelling and inference aspects. The course will include Gaussian process regression and classification as well as give examples of how Gaussian processes can be used as building blocks in more complex models. The participants will also be introduced to more recent advances in the field.
     
    The course will contain a mix of lectures, practical assignments and project work. The practical assignments will be based on pen & paper and the Python programming language. Other languages (such as Matlab and R) can be used, but it will require significantly more work from the participants.
     

    Session 1: Wednesday, 9 January, 10:15 » 12:00, 10:15 » 12:00, U358 (Otakaari 1)

    Introduction to the course and quick review of Bayesian inference and the properties of the multivariate normal distribution

    Session 2: Wednesday, 16 January, 10:15 » 12:00, U358 (Otakaari 1)

    Read Chapter 1 & 2 in Gaussian Processes for Machine Learning by Carl Rasmussen (http://www.gaussianprocess.org/gpml)

    Session 3: Wednesday, 23 January, 10:15 » 12:00, U358 (Otakaari 1)

    Read ch. 4.2 and ch. 5.1-5.4 in Gaussian processes for Machine Learning by Carl Rasmussen (http://www.gaussianprocess.org/gpml)

    Session 4: Wednesday, 30 January, 10:15 » 12:00, U358 (Otakaari 1)

    Read "Gaussian Processes for Big Data" by Hensman et al (http://www.auai.org/uai2013/prints/papers/244.pdf)

    Session 5: Wednesday, 6 February, 10:15 » 12:00, U358 (Otakaari 1)

    Read: "Gaussian process kernels for pattern discovery and extrapolation" (http://proceedings.mlr.press/v28/wilson13.pdf)

    Session 6: Wednesday, 13 February, 10:15 » 12:00, U358 (Otakaari 1)

    Glance through: "Spatiotemporal Learning via Infinite-Dimensional Bayesian Filtering and Smoothing" by Särkkä et al. (here)
    Get in the mood by watching: https://youtu.be/myCvUT3XGPc