An active memory as a model for information fusion

Abstract

Information fusion is a mandatory prerequisite for
\ncognitive vision systems. These are vision systems that apply rea-
\nsoning and learning on different levels of abstraction and corre-
\nspondingly have to deal with hypotheses from different categori-
\ncal domains. Following some principles of human cognition, we
\npresent an approach to information fusion that closely couples
\nreasoning and representation. We will discuss how processes like
\nprobabilistic contextual reasoning as well as functional and non-
\nfunctional requirements in storing data from different sources can
\nbe integrated by a unified XML based data representation. Due
\nto the interaction between active processes and data storage, we
\ncall our approach an active memory. Performance results of an
\nimplemented system as well as an evaluation of data fusion from
\ncontextual inference will be presented.