Mission

Information Fusion is a process of combining information collected from various disparate sources (multiple different sensors, data bases, knowledge bases, communication lines) into one coherent structure that can be used by a computer system to make a decision that leads to achieving the system's goal.
With the continuous progress in sensor and communications technologies the amount of information that is provided to computer systems is constantly increasing, overwhelming both the computer systems and the human operators. In spite of a significant progress in research on Information Fusion, there is still a lack of a theoretical framework for integrating disparate sources of information, and especially for developing (specifying, designing) such systems.

Our mission is to work towards a theoretical (formal) Information Fusion Framework in which the development of an Information Fusion System can be represented as an operation of combining theories associated with sensors, information sources, domains, system performance criteria, and the fusion process itself. The keystone of our approach is the recognition of the fact that fusion is an operation on those theories (information structures), rather than just on data items; it is developing new structures from old known structures and data.

We use category theory as the theoretical framework in which we represent, compose and transform information structures containing all knowledge in all of the phases of the fusion system development process. These structures are stored in reusable libraries (specifications). The development of the system is a process of refinement and transformation of system goals into a working system, using category theory operators (morphisms). This process preserves the system invariant expressed in terms of a performance criterion. In this framework, the resulting system is provably correct, i.e., it will obey the performance criterion, if the assumptions considered in the process are true.

Our goal is to establish a radically new approach to information fusion, resulting into a new discipline -- theory of information fusion. The basics of the information fusion theory developed in this project will establish solid grounds for further development and will point to directions for the search to solutions of fusion problems. Among others, we are working on:

• Foundations for incorporating performance measures and measures of effectiveness (MOE) of fusion, using all kinds of treatments of uncertainty
• Means for the rationalization of architectures of fusion and flexible architectures for the adaptation to changing system requirements
• Automatic tool support for designing Information Fusion Systems
• Basic theories of abstracting qualitative information from quantitative sensory data through the utilization of category theory and wavelet transformations