Examination arrangement

Course content

The course gives students the opportunity to implement many classic AI algorithms and use them as modules in large AI systems to perform tasks such as speech and image processing, simulated soccer (in the well-known Robocup on-line competition), Texas Hold'Em poker playing, and robot navigation. Some of the important AI methods that can appear in various projects include;

• the A* algorithm
• means-ends analysis
• decision-tree learning
• genetic algorithms
• neural networks
• bayesian classification
• case-based reasoning
• boosting and bagging

Through this work, students will gain an in-depth understanding of "AI in practice" as opposed to the combination of "AI in theory" and "AI on toy problems" that one experiences in the introductory and intermediate AI courses.

The course will consist of 2-4 projects, depending up the year and the extent of the individual projects. Each project will be supported by a series of lectures on relevant theoretical and practical issues surrounding the problem domain, while some class meetings will be reserved for interactive discussions of the problem and student progress.

Students will be free to program in the language of their choice, although Python, Java and C++ will be recommended.

Learning outcome

Students will gain hands-on experience designing and implementing relatively large AI projects. Students will gain valuable insights into why, when and how to use AI methods in realistic problems that they may encounter in their technical careers.

Learning methods and activities

50% standard lectures, and 50% interactive project discussions between students and teacher.

Students will be allowed to work alone or in groups of 2 (but no larger)

Further on evaluation

Course evaluation consists of 1-3 projects, which, together, consist of 6 components.

To pass the course, a student must receive a passing mark on at least 5 of the 6 components.

In the event of voluntary repetition, fail (F) or valid absence, the entire course must be retaken in a semester with teaching. All 6 components must be new submissions.

Specific conditions

Required previous knowledge

This course is only available for students admitted to the specialization in Artificial Intelligence in Computer Science (MTDT, MIDT), Informatics (MIT/MSIT) and Industrial Economics and Technology Management (MTIØT).

The course builds upon:

• TDT4120 Algorithms and Data Structures
• TDT4136 Introduction to Artificial Intelligence
• TDT4171 Artificial Intelligence Methods
• Requires previous knowledge in Discrete Mathematics comparable with MA0301 Elementary Discrete Mathematics.

Course materials

Lecture notes and complete project descriptions will be provided, as will any research articles of relevance to a project.For robotics projects, students will have access to a robot simulator and possibly to real robots (for a limited time).All materials are free.

Credit reductions