Automatic design of interpretable membership functions
Ralf Mikut, Jens Jäkel, Lutz Gröll · 2000
Most approaches for the data-based design of fuzzy rule-based classi ers (e. g. [1{4]) require a manual de nition of membership functions (MBFs). In large applications, this task is very diAEcult and time-consuming for human experts. As a practical application, a possible scenario for a manual design of membership functions in a medical expert system to support gait analysis is demonstrated (Section 2).