The Fuzzy Logic Systems (FLS) has been widely used to make up uncertainty and imprecision that are a part of the complex real world problems. However, the traditional fuzzy systems rely heavily on human experience to formulate rules as well as to calibrate membership functions hence limiting their flexibility and scalability. To overcome these drawbacks, machine learning (ML) algorithms have been combined with fuzzy logic resulting in intelligent and adaptive fuzzy systems. This paper provides an extensive overview of ML algorithms to be applied to a fuzzy logic system, including neuro-fuzzy models, evolutionary fuzzy systems, fuzzy clustering methods, and hybrid deep-learning-fuzzy systems. It is a critical review of system architectures, learning mechanisms, applications, benefits and shortcomings. Lastly, the existing difficulties and future research directions are outlined.