ABSTRACT
This chapter begins with some methodological aspects of fuzzy sets that are of paramount relevance in the context of pattern recognition. It presents information granulation, information granules, and elaborates Granular Computing on the concept of abstraction and its role in information processing. The chapter focuses on supervised learning with fuzzy sets by showing how several main categories of classifiers are constructed by taking into consideration granular information. It also discusses unsupervised learning and shows that fuzzy sets play a dominant role given the unsupervised character of the learning processes. Fuzzy sets offer an interesting option of quantifying available domain knowledge, giving rise to an idea of partial supervision or knowledge-based clustering. The chapter highlights the list of challenges and presents selected ideas of data and feature reduction, some of the possible formulations of the associated problems and look at their solutions. Fuzzy pattern recognition comes as a coherent and diversified setting of pattern recognition.
