More recently, fuzzy concepts have also been used in machine learning, giving birth to the field of fuzzy machine learning. This development has largely been triggered by the increasing popularity of machine learning as a key methodology of artificial intelligence (AI), modern information technology and the data sciences. Moreover, it has come along with a shift from knowedge-basedto data-drivenfuzzy modeling, i.e., from the manual design of fuzzy systems by human experts to the automatic construction of such systems by fitting (fuzzy) models to data.
In more classical applications like information processing and expert systems, fuzzy logic is primarily used for the purpose of knowledge representation, and inference is mostly of a deductivenature. Machine learning, on the other hand, is mainly concerned with inductiveinference, namely, the induction of general, idealized models from specific, empirical data. Thus, while the key importance of probability theory and statistics as mathematical foundations of machine learning is immediately understandable and indisputable, the role of fuzzy logic in this field is arguably much less obvious at first sight.