Flexibility and robustness of hierarchical fuzzy signature structures with perturbed input data
B. Sumudu U. Mendis, T.D. Gedeon, Lazlo Koczy · ANU Open Research (Australian National University) · 2006
We investigate the ability of fuzzy signature structure to cope with substantially reduced information as simulated by an experiment in which major branches were removed. We also investigated removal of significant number of individual data items. The results demonstrate that fuzzy signatures are robust under both of these conditions. This was our prediction, as fuzzy signatures are essentially hierarchical vector valued fuzzy sets designed for data sets and inconsistent substructures. To measure the succuss or otherwise of our experiments we introduced three new measures to compare results of two fuzzy signatures, being similarity, dissimilarity, and risk.