An Investigation into Fuzzy Clustering and Classification.
Michael Gray · Defense Technical Information Center (DTIC) · 1984
Pattern recognition algorithms based on fuzzy set theory were investigated and compared to their analogs which use traditional, or crisp set theory. The fuzzy K-means clustering algorithm was investigated and the fuzzy K-nearest neighbor and fuzzy 1-nearest prototype classifier algorithms were developed. These pattern recognition algorithms produce membership assignments (values from zero to one) for the samples considered. Thus, a sample's degree of belonging in a class can be assessed via these membership assignments.