The course will offer the student insight into state-of-the-art research on preferences in artificialintelligence. This will allow the students to gain a profound understanding of the relevance of preferences for a large variety of applications. They become familiar with some of the most advanced preference modeling techniques. Moreover, they learn to analyze related problems in different subfields of artificial intelligence and knowledge representation. This will increase their ability to detect similarities among problems, even if they are presented from very different perspectives.
Preferences are ubiquitous. They determine our decisions, from simple everyday decisions (such as having tea or coffee, ordering another glass of beer) to much more fundamental decisions (such as accepting a job offer, getting married, political participation). Preferences also play a tremendous role in many applications, such as logic programming, multi-agent systems, diagnosis etc. Given this, it is far from surprising that various subfields of artificial intelligence (AI) have come up with models for representing preferences and for reasoning and decision making based on these models. Whereas classical decision theory is based on a numerical representation of preferences (utilities), more recently qualitative as well as mixed qualitative/quantitative models of preferences have been a major focus of research.The course will present the most influential preference models developed in various subfields of AI. In particular, it will cover preferences in constraint reasoning, CP nets, preferences in nonmonotonic reasoning, in argumentation and in logic programming. Moreover, we will explore the challenges of joint decision making based on preferences, a central problem in multi-agent systems.
The course is based on two main parts. The first part will consist of 6 lectures which provide the necessary background and foundational material as well as an introduction to current research topics. In the second part, students have to apply the concepts and techniques presented in the lecture within a small project, which can either be concerned with a theoretical question or about implementation. Hereby, students will thus actively work with up to dateliterature and participate in current research conducted at our group.
ECTS breakdown: 3 ECTS = 75 HoursLecture 15hProject 60h
Not necessary
The course is for master and PhD students with background in formal logic and complexity theory. Some experience with knowledge representation and/or artificial intelligence will be helpful, but is not a necessary precondition for successful participation.
