Classification with fuzzy hypermatrices
Annamária R. Várkonyi-Kóczy, Balázs Tusor, János T. Tóth · 2016
This paper presents a novel classification method designed specifically for fast evaluation of problems with relatively low attribute quantities, such as image processing problems. The idea lies in the pre-calculation of fuzzy membership functions and their storage in an easily implementable structure. During the evaluation step the stored values can be accessed easily and quickly, making the processing of large quantities of data (like the pixels of images) significantly faster. A training algorithm is proposed and experimental results are shown to demonstrate the efficiency of the proposed technique.