IX – Neuro-fuzzy classification

Publisher Summary This chapter presents the applications of neuro-fuzzy classification to medical image compression and exploratory data analysis. It also reviews the basic concepts of fuzzy sets and the definitions needed for fuzzy clustering and presents several of the best-known fuzzy clustering algorithms and fuzzy learning vector quantization. The main difference between fuzzy and neural paradigms is that fuzzy set theory tries to mimic the human reasoning and thought process whereas neural networks attempt to emulate the architecture and information representation scheme of the human brain. It is therefore meaningful to integrate these two distinct paradigms by enhancing their individual capabilities in order to build a more intelligent processing system. This new processing paradigm is known under the name of neuro-fuzzy computing. The two basic components of fuzzy systems are fuzzy sets and operations on fuzzy sets. Fuzzy logic defines rules, based on combinations of fuzzy sets by these operations. The chapter also reviews the most important fuzzy clustering techniques and shows their relationship to non-fuzzy approaches. The main difference between traditional statistical classification techniques and fuzzy clustering techniques is that in the fuzzy approaches an input vector belongs simultaneously to more than one cluster while in statistical approaches it belongs exclusively to only one cluster. The chapter describes an adaptive fuzzy technique based on using different metrics to allow the detection of cluster shapes ranging from spherical to ellipsoidal clusters and an adaptive procedure similar to the generalized adaptive fuzzy n-means algorithm, but for shell prototypes.