ITI41820 Advanced Topics in Machine Learning (Spring 2022)

The course is connected to the following study programs

Elective course in the master programme in applied computer science full-time and part-time.

Recommended requirements

ITI41720 Machine Learning

Lecture Semester

Second semester (spring) in the full-time programme.

Second or fourth semester (spring) in the part-time programme.

The student's learning outcomes after completing the course

Knowledge

The student

• is the possibilities and advantages of employing the machine learning methods in the course as well as possible problems that may be encountered and how to overcome them.

• knows how the algorithms presented in the course work and their characteristics, for example which problems they work best for, overfitting, expected accuracy and computational requirements, for example how much benefit that accelerators may provide.

Skills

Given a machine learning application, the student is able to

• determine which theory and which methods that are presented in the course that are relevant and also how to apply them.

• perform hyperparameter tuning or in some cases even perform modifications of the source codes.

• use at least one implementation for each of the major machine learning techniques that are taught in the course.

General competence

The student

• is able to independently read machine learning papers and other literature and evaluate what works well and what does not for new problems.

• knows the terminology of machine learning and be familiar with the mathematics that is common in the field.

• knows the general behaviour of machine learning methods for example regarding how much data that is required, how to preprocess the data and ensure that its quality is sufficient.

Content

The course goes in depth on selected topics and methods within machine learning and their applications. Examples include:

• advanced neural net and deep learning models, such as: ResNET, Zero shot, GAN, LSTM.

• Evolutionary and bio-inspired algorithms algorithms (like GA, EA, ES, PSO, ACO, AIS) in search, optimization and classification.

• Program induction. Symbolic regression. Automatic programming.

• Markov models, Kernel methods. SVM

• Implementing machine learning in Industries and business

• Machine learning challenges and future

• Philosophical fundamental problems and ethical questions related to machine learning

The course syllabus will continuously be updated with methods from state-of-the-art research. Other topics may be chosen by machine learning group members each year and may vary depending on who is involved.

Forms of teaching and learning

The students will learn by attending seminars, papers and online material in the course reading list and above all by working on a project with a selected topic throughout the course and giving presentations at the seminars.

Workload

Approx. 280 hours.

Coursework requirements - conditions for taking the exam

The student must:

• give presentations at two seminars.

Coursework requirements must be accepted to qualify for the exam.

Examination

Project report and individual oral exam

The assessment is based on the project report and an individual oral exam. The project report is graded on the A - F grading scale. It is given a tentative grade of the report. This grade can be adjusted up to 2 stages at the oral exam.

The individual oral exam based on the course curriculum and project work. Approximately 30 minutes duration. No supporting materials allowed.

Examiners

External and internal examiner, or to internal examiners.

Conditions for resit/rescheduled exams

In case of re-examination, a new project must be carried out in agreement with the course instructor.

Course evaluation

This course is evaluated by a

• Mid-term evaluation (compulsory)

The responsible for the course compiles a report based on the feedback from the students and his/her own experience with the course. The report is discussed by the study quality committee of the faculty of Computer Sciences.

Literature

Last updated 22.10.2020. The reading list may be subject to changes before 1st of December 2021.

Online materials posted on the learning platform.