Нечіткі нейронні мережі

Abstract

The paper considers modern approaches to the design and application of fuzzy neural networks as hybrid artificial intelligence models that combine fuzzy logic theory and learning mechanisms of artificial neural networks. The main architectures of neuro-fuzzy systems are analyzed, including ANFIS models, hybrid neuro -fuzzy systems, logicoriented fuzzy neural networks, and deep fuzzy networks. The main application areas of fuzzy neural networks in classification, regression, forecasting, decision making, and intelligent control are discussed. It is shown that neurofuzzy models provide a balance between accuracy, adaptability, and interpretability, which is essential for modern intelligent systems.