Search

In this paper, we propose a new class of neuro-fuzzy systems. Moreover, we develop a novel method for reduction of such systems without the deterioration of their accuracy. The reduction algorithm gradually eliminates inputs, rules, antecedents, and the number of discretization points of integrals in the center of area defuzzification method. It then automatically detects and merges similar input and output fuzzy sets. Through computer simulations it is shown that accuracy of the system after reduction and merging has not deteriorated despite the fact that in some cases up to 54% of various parameters and 74% of inputs were eliminated. The reduction algorithm has been tested using well-known classification benchmarks. [ABSTRACT FROM AUTHOR]

Nanotechnology has opened numerous ways for physically realizing very sophisticated nanodevices that can be fabricated exclusively using molecular engineering methods. However, the synthesis procedures that lead to the production of nanodevices are usually complicated and time consuming. For this reason, the destination materials should be well designed. Therefore, numerical simulations can be invaluable. In this work, we present numerical simulations of the magnetic behaviour of magnetic units shaped into nanometric strips as a low dimensional layout that can be used as nano-systems of combinatorial logic. We showed that magnetic layouts that contain fewer than 16 magnetic units can take on a specific configuration as a response to the input magnetic field. Such configuration can be treated as an output binary word. The layouts that contained various numbers of magnetic units showed different switching characteristics (utterly different order of inverting of strips' magnetic moments), thus creating numerous combinations of the output binary words in response to the analog magnetic signal. The number of possible output binary words can be increased even more by adding parameters––the system's initial magnetic configuration. The physical realization of the model presented here can be used as a very simple and yet effective encryption device that is based on nanometric arrays of magnetic units rather than an integrated circuit. The same information, provided by the proposed system, can be utilized for the construction of a nano-sensor for measuring of magnetic field with the possibility of checking also the history of magnetization. [ABSTRACT FROM AUTHOR]