Till KTH:s startsida Till KTH:s startsida

Topics in Robotics, Module in Probabilistic Graphical Models

Logga in till din gruppwebb

Du är inte inloggad på KTH så innehållet är inte anpassat efter dina val.

This is a 3+ 3 credit module for the course Topics in robotics, and can be reported as (part of) any of DD3354, or DD3355. This means that you must contact the examiner of the Topics in robotics course in order to be registered and get credit for this module.

Summary:
This is an overview course of the state of the art in PGM. Students will become acquainted with historic and contemporary literature, as well as the most common problems.

Learning outcomes:
It is expected that after completing the course the student should be able to:

  • Have a good overview of the main types of PGM, and what applications they are used in. 
  • Understand how influence flows through these graphs.
  • Have an overview of methods used for inference and learning.
  • Implement and evaluate a state of the art PGM.

Course structure:

We will begin with four lectures summarizing the field, its problems, and relevant literature. Each lecture will be accompanied by a reading assignment expected to be done before the lecture.  Participation and discussion during lectures is required.  

All students will do at least 3 of the other student's tutorials and give feedback in the form of peer review.  Some of the tutorials can be from previous rounds of the course.  Tutorials will be assigned but based on the students preferences.

Those students wanting only 3 hp from the course should then improve one of the existing tutorials in some way.  

Then for those wanting 6 hp for the course: 

The students will then each choose an individual application, and related literature, with the assistance of the course teachers.   This might be for example based on one excellent journal paper and its background literature.

Students will each study their chosen topic in depth, implementing the techniques described and prepare a simple tutorial for the other students.  The tutorial should include a description of the problem and the theory.  It should also give the students a chance to try out the method on real data, for example in python or matlab.

The tutorials should  be revised after all feedback has come in and submitted as the final examination.

Contact: 

John Folkesson, johnf@kth.se

Administratörer