Methods in AI research, 2017-18

Methods in AI research

This is the official course page for Methods in AI research (MAIR). This page also accommodates the MAIR schedule. MAIR is an introductory course, and is meant to level out on the necessary prerequisites for the AI master at Utrecht University.

Motivation

AI is a field that is rooted in many disciplines. To pass the master, you will need to understand the values and basic principles of at least some of these disciplines. In this course, lecturers from three faculties (science, philosophy and psychology) will introduce you to three important disciplines in AI, viz. agents, reasoning, and cognitive processing. These three disciplines also form the tracks of the Utrecht AI master. Due to the highly multi-disciplinary character of AI, it may be difficult to relate the three track introductions. To remedy this, you will also work on an interdisciplinary project. This project is executed in teams and coordinated by staff from the cognitive processing and reasoning track.

Course goals

The course goals are to achieve knowledge and skills with respect to the following topics.

  1. The prerequisites of agent-oriented programming, in particular 2APL.
  2. The prerequisites of multi-agent learning, in particular Markov decision processes.
  3. The role of reasoning, in particular logic and linguistics, in AI. The ways in which reasoning can be formalized and of the main issues regarding such formalisations.
  4. The importance of transparency in automated decision making when researching and developing AI on which humans depend.
  5. The role of cognitive science, psychology, and linguistics in the field of artificial intelligence and vice versa, i.e., to understand fundamental concepts, methods, skills, and techniques from each of these fields to a depth that they can read basic scientific literature and gain further knowledge on these aspects in future courses, research, or independent study.
  6. To execute an interdisciplinary project in the field of AI by using methods from the domains of reasoning and cognitive processing and how to report the findings in writing and orally.
  7. To learn new concepts, knowledge and/or skills from other domains in AI through interaction with students that have a different background.

Course material

Due to the breadth of topics and its large team of contributors it is inevitable that the course material is distributed.

Attendance

Attendance is recommended but not obliged. Past experience has learned that student attendance and effort (e.g., completing assignments, reading papers) highly correlates with the success in passing the course. Attendance lists may be distributed during classes to have an up-to-date inventory of who attends.

Examination

Examination is per track and project.

Exam formWeight of final grade
Agents trackwritten exam5/15
Reasoning trackwritten exam and homework3/15
Cognitive Processing trackwritten exam3/15
Projectreport and final product4/15

You will pass the course only if all course components are completed and graded, and if the final raw average is at least 5.5. You can take part in a resit if you missed an exam for urgent reasons or if you had a grade lower than 5.5 on that specific exam. For written exams, one resit opportunity per track is available. For the project a special resit opportunity arrangement exists, see the project manual on blackboard.

If you have special requirements due to a physical or mental disability, contact the study advisor of AI, Corine de Gee, as soon as possible. MAIR has mid-term exams so we need to be informed about possible handicaps right away. As soon as your application is granted you should inform all coordinators timely by e-mail, at least three weeks before the first exam (scheduled mid course), with Corine de Gee in the CC. Lecturers don't need to know why you have a special requirement, but do need to know what extra conditions are granted by the study advisor. We will do our best to accommodate.

Fraud and plagiarism

The regulations and considerations of what constitutes fraud and plagiarism are defined in the Graduate School of Natural Sciences Master’s Degree Programme Education and Examination Regulations (OER). Detection of a violation may lead to being excluded from examination and even being removed from the master’s program.

Contact information

Overall course and agents track Projects and cognitive processing track Reasoning track
Dr. Gerard Vreeswijk
Dept. of computer science
Dr. Chris Janssen
Dept. of experimental psychology
Dr. Rosalie Iemhoff
Dept. of theoretical philosophy

Contact the track coordinator if you have a question about a specific track. Otherwise, contact the course coordinator. We prefer to answer questions in class or during breakes rather than by mail. Thanks.

Outside MAIR, the AI master programme track coordinators are Dr. Mehdi Dastani and Dr. Frank Dignum (Agents); Dr. Rick Nouwen and Dr. Rosalie Iemhoff (Reasoning); Dr. Chris Janssen (Cognitive Processing).


This page last modified at Thu, 07 Sep 2017 10:25:35 +02001.