Evolutionary Fuzzy Systems: A Survey of Hybrid Intelligence for Optimization and Decision-Making

Abstract

Evolutionary Fuzzy Systems (EFSs) is a combination of evolutionary algorithms and fuzzy logic systems, providing the ability to globally optimally solve a problem with the interpretability and human reasoning of fuzzy systems. The survey examines the theory, hybridization approaches, and applications of EFSs in a wide range of fields, such as in control systems, robotics, healthcare, data mining, optimization, and smart systems. We present several important elements including the membership function tuning, rule base optimization and structural adaptation that is possible with the help of evolutionary algorithms, and their potential to automatically design and optimize fuzzy systems. Other issues explored by the survey are performance evaluation methodologies, benchmarking practices and scalability, computational complexity, trade-offs between interpretability and accuracy, and real-time adaptability. Moreover, it outlines the new research directions, such as explainable evolutionary fuzzy systems, their combination with deep learning and reinforcement learning, multi-objective optimization, and their implementation in edge and IoT settings. By consolidating of the state-of-the-art advances, this survey can give some insights into the possibilities and limitations of EFSs, and it can be concluded how future research should proceed in creating adaptive, robust, and interpretable intelligent systems. In general, EFSs still hold great potential in tackling complex, uncertain and high dimensional real world problems.