acad. prof. dr. Ivan Bratko
Faculty of Computer and Information Science, University of Ljubljana
Title of the invited lecture: Robot Learning and Planning with Qualitative Representations
To execute tasks in an unknown environment, a robot has to learn a model of the environment and use the learned model for task planning. One approach to this is reinforcement learning with deep neural networks. However, a drawback of this approach is lack of comprehensibility. In the interest of Explainable AI, the use of qualitative representations is more promising. In this talk, an approach based on ideas of qualitative modelling and simulation will be presented. The approach will be illustrated on the problems of learning to fly a quadcopter, and a humanoid robot learning to walk.
Short CV: Ivan Bratko is professor of computer science at the Faculty of Computer and Information Science, Ljubljana University, Slovenia. He has conducted research in several areas of Artificial Intelligence, including machine learning and data mining, knowledge-based systems, intelligent robotics, qualitative modelling and heuristic search. His main interests in machine learning have been in learning from imperfect data, Inductive Logic Programming, combining learning and qualitative reasoning, autonomous robot discovery, and various applications of machine learning including medicine, ecological modelling and control of dynamic systems. Ivan Bratko is the author of widely used text Prolog Programming for Artificial Intelligence (4th edition: Pearson Education 2012) and over 200 scientific publications. He is Fellow of EurAI, member of SASA (Slovene Academy of Sciences and Arts) and Academia Europaea.