On Adaptive Open-World Modeling Based on Information Fusion and Inductive Inference

Technical report


Achim Kuwertz


Technical report IES-2010-16. In: Jürgen Beyerer, Marco Huber (eds.), Proceedings of the 2010 Joint Workshop of Fraunhofer IOSB and Institute for Anthropomatics, Vision and Fusion Laboratory, Karlsruher Schriften zur Anthropomatik, vol. 7, KIT Scientific Publishing, 2010.





In this technical report, a conception for adaptive open-world modeling for cognitive information systems is presented. In cognitive systems, a world model serves as information storage for sensor data and thus represents an abstract, simplified copy of the observed environment. In order to allow for a high-level information processing on a semantic layer, the represented objects are backed by a semantically enriched domain model containing a priori knowledge. Such prior knowledge generally contains only a fixed number of object concepts, thus constituting a closed-world model. However, in many real-life applications, the considered environment is not closed. For coping with changing environments, a cognitive system must be equipped with an adaptive world model able to adjust to an observed open environment. For designing such an open-world model, this report evaluates and summarizes information fusion and concept learning techniques.