Neural net — Fuzzy logic rules mapping for dynamic of fuzzy sets boundaries

Beauty, quality, performance, shape, or form are just a few characteristics that are hard to quantify. Expecting a computer to deal with such ambiguous properties is complicated since even humans sometimes have trouble agreeing on their meanings. L. A. Zadeh in 1965 at the University of California at Berkeley introduced the concept of “Fuzzy Sets”. It is not the intent of the author of this paper to evaluate Fuzzy Logic as a whole due to its broadness. Rather by analyzing a few characteristics about fuzzy logic considered weaknesses, the author wishes to provide information about current solutions as well as offer other innovations. It is well known that if a fuzzy system is tweaked optimally and assigned set values to the truths, then the system ceases to become fuzzy. In other words, the fuzzy system lacks adaptability. The possibility of using neural networks to adjust fuzzy logic sets is studied. The results are accomlished by comparing three different systems: one with fuzzy logic only, one with neural networks only, and finally one with a combination of fuzzy logic and neural networks. The basis of the problem on all three models involves balancing a weight on an inverted pendulum balance.