Neuro-Fuzzy Systems and their Applications

Summary form only given. In the lecture, we incorporate various flexibility parameters to the construction of neuro-fuzzy systems. This approach dramatically improves their performance, allowing the systems to perfectly represent the pattern encoded in data. It should be noted that the lecture provides a framework for the unification, construction and development of neuro-fuzzy systems. It presents complete algorithms in a systematic and structured fashion, easing the understanding and implementation. The strength of the lecture is that it provides tools for possible applications in business and economics, medicine and bioengineering, automatic control, robotics, decision theory and expert systems. In the lecture, we discuss a new class of neuro-fuzzy systems characterized by automatic determination of a fuzzy inference (Mamdani/ logical) in the process of learning. Consequently, the structure of the system is determined in the process of learning. In addition to automatic determination of a system type, we introduce several flexibility concepts in the design of neuro-fuzzy systems: softness to fuzzy implication operators, to the aggregation of rules and to the connectives of antecedents; certainty weights to the aggregation of rules and to the connectives of antecedents; parameterized families of t-norms and t-conorms to fuzzy implication operators, to the aggregation of rules and to the connectives of antecedents.