A FUZZY COMPUTATIONAL MODEL IMPLEMENTATION
Kostadin Kratchanov, Ivan Stanev, Desislava S. Sabotinova · Czech digital mathematics library · 1992
During recent years we developed the foundations of the theory of a kind of rule-based systems which we called controlled rule-based systems (CRBS) (e.g. [1] [3]), and gained some experience in applying this approach in building instrumental tools and particular systems (earlier works cited in [4]). In fact, we consider the CRBS's as a convenient universal computational model, capable of representing any particular algorithm (see also [5]). Moreover, the whole theory developed so far deals with fuzzy CRBS's (FCRBS) actually. As stated as early as in [6], fuzzification is possible in many points, and equivalent transformations may be considered. The theoretical model allows arbitrary fuzzifications but in any particular case only few of these opportunities are really necessary. One has also to keep in mind that any fuzzification yields big losses of memory and computational time. For those reasons in [2], where algorithms of the interpreters in FCRBS's have been considered, the following line of exposition has been adopted. Firstly, the general case is studied when all the composite fuzzy relations of the behaviour are essentially fuzzy. Secondly, a universal algorithm is given when some of these composites may be in fact partial or total functions. Finally, based on this general but non-effective algorithm, one can easily form efficient special algorithms for any particular case.