An fMUX architecture: data modularization and mixed-mode system modeling

Abstract

 In this study, we introduce a concept of a fuzzy multiplexer (fMUX) generalizing a well-known idea of multiplexers existing in two-valued logic and digital systems. A generic topology of the fMUX is presented and its functional behavior studied. We show that a design process with fMUXs is concerned with their supervised learning. A detailed learning scheme is discussed. Numerical illustration is included. It is shown that fMUX support an idea of mixed-mode modeling in which we encounter data-information granules quite often exhibiting variable granularity and this diversity lends itself to a modular way of model development.