PSY9511 – Machine learning

The field of machine learning is devoted to building methods that identify patterns in data, in order to make automated predictions or decisions.

Machine learning is becoming an increasingly important method in Psychology as datasets increase in size and complexity, e.g. data from registers or brain imaging. The goal of this course is to give Ph.d. students an introduction to machine learning that will enable them to apply these tools to their own research, and with which to further navigate the literature. The lectures will cover core concepts and give an overview of different methods suitable to different research problems. Through practical exercises, students will gain an introduction to important software packages like keras and xgboost. The students can choose between using R or Python for the practical exercises.

PSY9511 Machine learning is the PhD-level version of PSY4319 Machine learning, an elective course within the master's program. The content, schedule, and reading list for PSY9511 Machine learning are the same as for PSY4319 Machine learning.

Knowledge

After completing the course, students will have knowledge of:

• The main concepts of machine learning, including supervised learning, unsupervised learning, and reinforcement learning.

• The difference between regression and classification, and have basic familiarity with core classification methods such as discriminant analysis, naïve bayes, and logistic regression.

• The problem of overfitting, and the common methods for model evaluation such as cross-validation, training, test, and validation sets.

• Machine learning methods for variable selection and dimension reduction.

• Methods for handling nonlinear relationships in data.

• Tree based methods (random forests, gradient boosting).

• Basic deep learning methods for image and text data.

• How machine learning differs from classical statistics.

• Unsupervised methods such as clustering.

Skills:

After completing the course, students will be able to:

• Apply machine learning models to address new research questions.

• Implement machine learning models using appropriate software.

• Evaluate models using techniques such as cross-validation.

There is a limited number of seats due to joint teaching with the master’s level version of the course, PSY4319 Machine learning.

PhD candidates at the Department of Psychology and master students within the field of Cognitive Neuroscience will be given priority, but it is also possible for others to apply for the course.

Candidates admitted to a PhD-program at the Department of Psychology can apply in StudentWeb.

Candidates external to the Department of Psychology program can apply by filling out and sending in an electronic registration form. Applicants receive an email shortly after the deadline if admitted to the course.

All candidates need to be formally registered for the course before the first day of teaching. 

The teaching will be organized into eight, three-hour long seminars. The seminars consist of a mix of lectures introducing the day’s topic, and work on practical exercises supervised by the teacher.

The course has joint teaching with the master course PSY4319 Machine learning.

In order to qualify for the exam 6 out of the 8 practical exercises must be approved in Canvas, in addition to the candidate attending a minimum of 6 out of the 8 seminars (75% of the teaching).

The exam consists of a 3-hour written examination at the university exam facilities at Silurveien.