Neuro-fuzzy ensembles: A systematic mapping study

In the last decade, neurofuzzy networks have received considerable attention from academia. These systems strike a tradeoff between the performance of artificial neural networks and the interpretability of fuzzy inference systems expressed through fuzzy rules. Many researchers, however, are still dissatisfied with the performance of single neuro-fuzzy systems and are constantly working to improve them using ensemble techniques. This study conducts a systematic mapping of relevant studies on the application of ensemble techniques to neuro-fuzzy systems. Many aspects of the study are highlighted, including publication years, sources, contribution types, and application domains. As result, 48 articles published from 2000 to Mars 2022 were selected from six digital libraries (Science Direct, IEEE Xplore, ACM Digital Library, PubMed, Wiley, and Google Scholar. The findings revealed that the number of studies employing ensemble techniques with neuro-fuzzy systems is low and unstable. As the most used neuro-fuzzy system, ANFIS, from the Takagi-Sugeno-Kang category, is being studied in a wide range of application domains, most notably finance, medicine, and ecology.