Repository of DUTH

Decision making and Fuzzy Logic (Doctoral thesis)

Βλάμου, Ελένη/ Vlamou, Eleni

Fuzzy logic theory provides improved solutions to problems with a high degree of uncertainty and it is used in several scientific and research areas. This dissertation focuses on the perception of fuzzy logic structure, fuzzy logic rules, and systems. Fuzzy sets theory is presented as well as fuzzy inference systems which are analyzed in depth in order to perceive fuzzy logic application effectiveness. In particular, artificial neural networks and fuzzy logic systems blending allow the development of adaptive systems, which are trained with the use of algorithms like such as Back-Propagation algorithm in order to improve the fuzzy neural network performance. The improved efficiency of fuzzy neural networks is confirmed with Back Propagation algorithm implementation in Matlab. Additionally, the implementation of fuzzy logic systems in order to analyze brain signal within the frame of pattern recognition is discussed. The description of linear pattern recognition methods for brain signal analysis (such as Fast Fourier transforms, Wavelet transformation and Vector Quantization) is presented in order to confirm the fuzzy neural networks (SOMF), the fuzzy classifiers and adaptive neuro-fuzzy inference system classifiers (ANFIS) efficiency. Furthermore, the use of fuzzy logic networks implementation within the frame of epidemiology diagnosis is presented. Therefore, several epidemiological models (such as stochastic epidemiological models, e.g. SI and SIS, and deterministic epidemiological models such as the SIR model) are analyzed in order to support fuzzy logic networks superiority. Thus, fuzzy logic SI and SIS models analysis additionally support the efficiency of fuzzy logic systems in the case of decision making in the field of epidemiology. Finally, fuzzy logic systems implementation is presented in order to confirm fuzzy genetic algorithms enhanced performance. An analysis of the basic principles and characteristics of genetic algorithms as well as the genetic factors involved are presented. Within this frame, the enhanced performance of fuzzy systems and fuzzy genetic algorithms efficiency are supported.