FIK3616 Learning Machines 7.5 credits

• Foundations of AI for learning machines. (Default lecturer first year Magnus Boman)
• History of learning machines. (Nina Wormbs)
• The future of learning machines. (Magnus Boman)
• TBC. (Anders Holst)
• Pronouncers. (Magnus Boman)
• Multi-AI (AI2AI) systems. (Magnus Boman)
• Concept formation in learning machines. (Daniel Gillblad)
• Deep learning. (John Ardelius)
• Systemic properties of large-scale learning machines. (Daniel Gillblad & Magnus Boman)
• Critical perspectives and fear of learning machines. (Francis Lee)
• Massive data for learning machines. (Jim Dowling)
• Applications of learning machines. (Magnus Sahlgren & Jussi Karlgren)
• Learning from failure in combinatorial problem solving. (Christian Schulte)

1. Autonomously solving problems
   Applying existing as well as future tools to building LMs
   Self-testing understanding and critiquing
   Interpreting the work of others
2. Mastering abstraction
   Recognising what an LM is (not)
   Identifying relevant concepts and applicable methods/tools
   Mastering the meta-level, modelling LMs
   Associating different relevant concepts with LMs
   Instrumentalising abstract concepts relevant to LMs
3. Implementing LMs
   Using tools in the LM context
   Exploring the effects of assumptions on a concept
   Programming (and testing) LMs
   Assessing the adequacy and complexity of LM programs

Ph.D students and master students planning to enroll on a Ph.D program.

Recommended Prerequisites:

Discrete mathematics, linear algebra, machine learning, programming, AI.

Relevant articles and research papers, plus documentation from Internet sources. During the course, a compendium will be developed, with all of the lecturers (and possibly their collaborators or students) as invited contributors. Students will be motivated to comment on, and influence the contents of, the compendium. Such influence can come in the form of course examination. The outcome will not be a collection of individual chapters by individual authors, but rather a monograph with many co-authors.

Some default lecturers have already suggested literature to cover their respective lectures.

Example (Lee):

1. Solon Barocas, Sophie Hood, Malte Ziewitz. (2013) Governing Algorithms: A Provocation Piece http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2245322
2. Ziewitz, M. (2011). How to think about an algorithm? Notes from a not quite random walk. Discussion paper for Symposium on "Knowledge Machines between Freedom and Control", 29 September 2011.

If the course is discontinued, students may request to be examined during the following two academic years.

• EXA1 - Examination, 7.5 credits, grading scale: P, F

Based on recommendation from KTH’s coordinator for disabilities, the examiner will decide how to adapt an examination for students with documented disability.

The examiner may apply another examination format when re-examining individual students.

Flexible exam: essay, documented program, contribution to course compendium, documented applied work at company, All types of exam have the same deadline.

Completed exam and at least two thirds of the lectures attended.

Magnus Boman

• All members of a group are responsible for the group's work.
• In any assessment, every student shall honestly disclose any help received and sources used.
• In an oral assessment, every student shall be able to present and answer questions about the entire assignment and solution.