Dealing with uncertainty in AI: an introductory discussion

Abstract

In this seminar, I will discuss the nature of the uncertainty that we find in our AI problems, and some of the mathematical and computational methods that we can use to deal with it. The discussion will be broad but informal. The viewpoint that I hope to convey is that there are many fundamentally different types of uncertainty, and that different methods are needed to deal with these different types in an adequate way. I will consider both probabilistic and non-probabilistic methods, including fuzzy sets, possibility theory, Dempster-Shafer theory, and qualitative approaches. This introductory seminar marks the beginning of my sabbatical stay at Trinity College Dublin, whose aim is to study the above topics. It will be followed by a series of more technical tutorials on non-probabilistic methods to deal with uncertainty. The aim of this first seminar is to start a discussion with any interested partners at TCD and TUD. Specifically, I would like to discuss local projects from the point of view of uncertainty: what uncertainty is present, how it is being dealt with (or not) currently, and if it could be dealt with in a better way.

Short Bio

Alessandro Saffiotti is full professor of Computer Science at Örebro University, Sweden, where he leads the AASS Cognitive Robotic Systems Lab. He is also adjunct professor at the University of Bologna, Italy. He has been active since 1985 in the field of Artificial Intelligence, especially in the areas of planning, knowledge representation and reasoning under uncertainty, and in application domains such as social robotics and technology for elderly people. His main scientific drive has been the integration of AI and Robotics: you may say "how to give a brain to a body, or a body to a brain"! He has published more than 200 scientific papers, which received close to 12,000 citations for an h-index of 56 (Google Scholar). He served as expert advisor for the European Commission, as director of euRobotics, the European robotics association, and as scientific manager of AI4EU, the bootstrap project to develop a European AI platform and ecosystem. He participated in 18 EU projects, several EU networks, and many national projects. He has been on the editorial board of Artificial Intelligence and of the International Journal of Social Robotics. He is a member of AAAI, a senior member of IEEE, and a EurAI fellow. More information: Alessandro Saffiotti's Örebro University profile

Venue

Large Conference Room, O'Reilly