Intelligent learning techniques for multi-source information fusion environments

The paper delineates the scope and potential of intelligent learning techniques in the design and development of information fusion processors for deployment in multisource environments. First, the need for nontraditional distance metric concepts capable of handling broad spectrum of feature information sources that may be input to a features in-decision out (FEI-DEO) type of fusion process is brought out. The potential of recently reported advances in the artificial intelligence community in meeting these requirements is discussed. This is followed by a discussion of the role of intelligent fusion learning techniques in the context of decisions in-decision out (DEI-DEO) type of fusion processes. This includes a brief presentation of a newly defined measure of performance of such DEI-DEO fusion processes. The role of intelligent learning techniques in facilitating the development of self-improving fusion systems, is addressed next. Finally, the learning in the context of features in-features out (FEI-FEO), data in-data out (DAI-DAO) and data in-features out (DAI-FEO) modes of fusion are discussed briefly to cover the spectrum of possible fusion modes