Deep Neuro-Fuzzy System application trends, challenges, and future perspectives: a systematic survey

Area of Science:

• Artificial Intelligence
• Machine Learning
• Computational Intelligence

Background:

• Deep neural networks (DNNs) offer powerful data processing but lack transparency.
• The 'black-box' nature of DNNs motivates hybrid approaches for enhanced interpretability.
• Deep neuro-fuzzy systems (DNFS) emerge as a solution, integrating DNNs with fuzzy logic.

Purpose of the Study:

• To conduct a systematic review of Deep Neuro-Fuzzy Systems (DNFS) research.
• To analyze progress, trends, challenges, and future scope in the DNFS domain.
• To provide a comprehensive overview of DNFS implementation and application.

Main Methods:

• Systematic literature review using a study mapping process.
• Search conducted across five scientific directories for publications from 2015-2020.
• Analysis of 105 identified studies to address specific research questions.

Main Results:

• Identified and analyzed 105 DNFS studies published between 2015 and 2020.
• Mapped the concept, optimization methods, research intensity, and application domains of DNFS.
• Confirmed a rapidly growing research trend in DNFS with significant future potential.

Conclusions:

• DNFS research is expanding rapidly, offering high interpretability and accuracy.
• The review highlights key trends, research gaps, and challenges in the DNFS field.
• Provides guidance for future research, fostering advancements in DNFS techniques and applications.