Fuzzy-Rough Data Mining

It is estimated that every 20 months or so the amount of information in the world doubles. In the same way, tools that mine knowledge from data must develop to combat this growth. Such techniques must be able to extract useful, meaningful patterns efficiently in the presence of large amounts of noisy, redundant and sometimes misleading information. To be able to achieve this, data mining is often reliant upon suitable data cleaning and reduction processes to allow a better quality of knowledge to be discovered. Indeed, the computational complexity of some techniques prohibits their application to large volumes of data, and the reduction step is a necessity. Fuzzy-rough set theory provides a framework for developing such applications in a way that combines the best properties of fuzzy sets and rough sets, in order to handle uncertainty. As will be discussed later, fuzzy sets and rough sets both model different aspects of uncertainty encountered in the real world. Their combination can therefore result in powerful methods for data mining and reduction. This chapter presents a general overview of recent fuzzy-rough data mining developments, including feature selection and classifier learning.