Machine Learning DD2431

Welcome to this course in Machine Learning, DD2431, Oct-Dec 2007.

How can a computer program automatically improve its behavior based on previous experience? Such questions often come up when designing game-playing programs, but also in robotics or when designing adaptive man-machine interfaces. The subject touches on artificial intelligence, statistics, information theory, biology and control theory. The goal of this course is go give basic knowledge about the theories and algorithms commonly used in the area.

News

If you have remaining labs to be examined, contact Jenny and/or Simon.

In "Mina Sidor", your result on the exam will only show up as a "P" (Pass) or "F" (Fail). This is due to an error in the system that we are probably not authorized to correct. The correct grade will show up as the final grade for the course once all your labs are finished and reported.

The results from the written examination on Dec 20 are ready. You can see your result via the res command. The result Fx means that you made errors on the A section but may still pass the exam after a small extra examination ("komplettering"). Contact Örjan for a suitable time.

Solution for the exam is available here.

Faculty

• Course leader: Örjan Ekeberg
• Lecturers: Örjan Ekeberg and Hedvig Kjellström

Literature

Textbook:

Mitchell: Machine Learning
McGraw-Hill, 1997
ISBN: 0-07-115467-1

Article on Adaboost:

R.E. Schapire. A brief introduction to boosting. In Proceedings of the Sixteenth International Joint Conference on Artificial Intelligence, 1999.

Article on Bagging:

J.R. Quinlan. Bagging, boosting, and C4.5 In Proceedings of the Thirteenth National Conference on Artificial Intelligence and the Eighth Innovative Applications of Artificial Intelligence Conference, pages 725-730, Menlo Park, August 4-8 1996, AAAI Press / MIT Press

Teaching

Teaching is in the form of lectures plus instructions in connection with the lab assignments. All lectures are in English.

• Schedule and questions of interest for the lectures.

Slides from the lectures

Introduction	Printable	Viewable
Concept Learning	Printable	Viewable
Decision Trees	Printable	Viewable
Artificial Neural Networks	Printable	Viewable
Bayesian Learning	Printable	Viewable
Bagging and Boosting	Printable	Viewable
Instance based Learning	Printable	Viewable
Reinforcement Learning	Printable	Viewable
Genetic Algorithms	Printable	Viewable
Learning Theory	Printable	Viewable
Rule based Learning	Printable	Viewable
Summary	Printable	Viewable

Lab Instructions

• Instructions for Lab 1 (Concept Learning)
• Instructions for Lab 2 (Bayes Classifiers and Boosing)
• Debugging help for Lab 2
• Instructions for Lab 3 (Reinforcement Learning)
• Instructions for Lab 4 (Genetic Algorithms)

Course Examination

• Written Exam (Tentamen)
• Four Lab Assignments

You do not have to register separately for the written exam.

Last years exam is available here:

• Last years exam in Swedish and English.
• Corresponding solution (Swedish only, sorry).