Not less importantly, the fuzzy community needs to get out of its isolation. This problem does not seem to be limited to machine learning but similarly applies to other application domains. There is a tendency to create a closed fuzzy X community (fuzzy control, fuzzy databases, fuzzy image processing, etc.) in parallel to the core X community. It is completely legitimate to have journals and conferences specifically devoted to fuzzy sets, just like there are journals and conferences on, say, probability theory and statistics. However, the enormous success of statistics is due to the fact that it has been established as a methodological basis in many application domains: the existence of mathematical statistics is largely justified by the multitude of applied statistics.
There is no doubt that fuzzy logic has a role to play, too, not only in mathematics, computational sciences and engineering, but in science in general. Yet, in addition to the fundamental problem and mammoth task of teaching people to think in terms of “partial truth” (a true paradigm shift in the history of science), claiming this role requires applications that show the merits of fuzzy logic in a convincing way. There are many research fields willing to ac-cept pragmatic arguments of that kind. However, researchers from these fields are not part of the fuzzy community, and typically even unaware of fuzzy logic. Therefore, we cannot expect them to adopt fuzzy logic methodology by themselves, let alone to attend fuzzy logic conferences to learn about our tools. Instead, it is our task to make them aware of fuzzy logic and convince them of its advantages—at least if we want fuzzy logic to survive next to other mathematical an AI methodologies in the long run.