Pattern Recognition, 4th Edition

14.4. Possibilistic Clustering

The algorithms of this section are relaxed from constraints such as in (14.17) and (14.65) [Kris 93, Kris 96]. Speaking in the terms of Section 14.3.5, this means that the vector y, with coordinates the u^ij's, will be allowed to move anywhere in the H^m hypercube, that is,and(14.66)This change in the constraints has an important impact on the interpretation of the u^ij's. In the fuzzy framework, u^ij denotes the grade of membership of xⁱ in the jth cluster. Here, u^ij may be interpreted as the degree of compatibility ...