Exploration of Fuzzy System With Applications

2. Fuzzy Data Mining

With the development of electronic data processing, more and more data were available electronically. This led to the situation in which the masses of data, for instance in data warehouses, exceeded the human capabilities to recognize important structures in these data. Classical methods to ‘data mine’, such as cluster techniques, and so on, were available, but often they did not match the needs. Cluster techniques, for instance, assumed that data could be subdivided crisply into clusters, which did not fit the structures that existed in reality.

Fuzzy set theory seemed to offer good opportunities to improve existing concepts. Bezdek was one of the first, who developed fuzzy cluster methods with the goals, to search for structure in data to reduce complexity and to provide input for control and decision making. Different approaches have been followed: hierarchical approaches, semi-formal heuristic approaches, and objective function clustering. Because the area of (intelligent) data mining becomes more and more important, the development of further fuzzy methods in this area can be expected.

A lot of work has already been done to explore the utility of fuzzy set theory in various subareas of systems analysis. However, the subject of systems analysis is too expensive to be covered here in a comprehensive fashion. Hence, we can cover only a few representative aspects and rely primarily on the Notes at the end of this chapter to overview the rapidly growing literature on this subject. The most successful application area of fuzzy systems has undoubtedly been the area of fuzzy control. It is thus appropriate to cover fuzzy control in greater detail than other topics. Our presentation of fuzzy control includes a discussion of the connection between fuzzy controllers and neural networks, the importance of which has increasingly been recognized. Furthermore, we also discuss the issue of fuzzifying neural networks, Genetic Algorithms.