IN9310 – Advanced Deep Learning for Image Analysis
Course description
Schedule, syllabus and examination date
Course content
This course presents advanced methods in deep learning to work with image data, covering key advanced deep learning algorithms and concepts for training neural networks. This course will complement the introductory concepts presented in IN4310, and will build practical skills on application of deep learning to image data. The course’s focus is on advanced learning and representations techniques for image analysis and understanding, such as visual transformers, graph-based models, adversarial networks, diffusion models, recurrent networks, among others. The course will also cover self-, semi-, and unsupervised learning regimes.?
Learning outcome
After this course you will be able to:?
- create a project plan and develop it to solve problems based on advanced deep neural networks with image data;
- have insight about novel (state-of-the-art) methods in Deep Learning used with image data and their relationship with prior work;
- create (design and train) advanced neural networks from scratch and pre-trained models, and finetune the neural networks;
- judge different network architectures designs (for example, adversarial networks, diffusion models, attention-based models, graph-based models, among others), their training regimes (supervised, semi-supervised, un-supervised and self-supervised learning), and in what contexts they are suitable;
- differentiate among supervised, semi-supervised, un-supervised and self-supervised learning regimes;
- create neural networks using standard libraries, for example, Pytorch;
- create (design and write) a scientific paper presenting the results of the research done.
The PhD-variant will also look at selected new research articles within deep learning.
Admission to the course
PhD candidates from the University of Oslo should apply for classes and register for examinations through?Studentweb. If a course has limited intake capacity, priority will be given to PhD candidates who follow an individual education plan where this particular course is included.
Recommended previous knowledge
Students ought to have taken IN4310 – Deep Learning for Image Analysis prior to taking this course.
Overlapping courses
- 10 credits overlap with IN5310 – Advanced Deep Learning for Image Analysis.
Teaching
2 hours of lectures, discussion, and learning activities as well as 2 hours of practical exercises per week.?
The course will have mandatory assignments as well as a project. The project work is done in small groups during the semester along with the other activities, and guidance will be given during the practical exercises. As such, students must attend the lectures and the practical exercises to gain the full benefit of the teachings.
The project will be to?prepare and delimit a problem, propose a solution, develop it, and present the results. (The problem and solution must be approved before developing the project.) The results must be presented to the peers during the semester and through a paper draft describing the project and its results.
The PhD-variant will have an extended syllabus compared to the main?course?and the project will be more extensive than for?the main?course.
Examination
The evaluation will be a group presentation and a final written report for the groups project, which counts 100% towards the final grade. The presentation and the report have an individual component where the students must demonstrate their contribution. Students will be given individual grades.
All mandatory assignments must be approved?prior to the exam.
It will also be counted as one of?your three?attempts to sit the exam for this course, if you sit the exam for one of the following courses: IN5310 – Advanced Deep Learning for Image Analysis
Examination support material
No examination support material is allowed.?
Grading scale
Grades are awarded on a pass/fail scale. Read more about?the grading system.
More about examinations at UiO
- Use of sources and citations
- Special exam arrangements due to individual needs
- Withdrawal from an exam
- Illness at exams / postponed exams
- Explanation of grades and appeals
- Resitting an exam
- Cheating/attempted cheating
You will find further guides and resources at the web page on examinations at UiO.