Deep Neural Fuzzy System Oriented toward High-Dimensional Data and Interpretable Artificial Intelligence

Fuzzy systems (FSs) are popular and interpretable machine learning methods, represented by the adaptive neuro-fuzzy inference system (ANFIS). However, they have difficulty dealing with high-dimensional data due to curse of dimensionality. To effectively handle ensure optimal performance, this paper presents a deep neural fuzzy (DNFS) based on subtractive clustering-based ANFIS (SC-ANFIS). Inspired learning, SC-ANFIS is proposed adopted as submodule construct DNFS in bottom-up way. Through ensemble hierarchical submodules, can not only achieve faster convergence, but also complete computation reasonable time high accuracy interpretability. By adjusting structure parameters DNFS, performance be improved further. This performed profound study combination inputs for DNFS. Experimental results five regression datasets various dimensionality demonstrated that solve dimensionality, higher accuracy, less complexity, better interpretability than previous FSs. The superiority validated over other recent algorithms especially when higher. Furthermore, built each two shared between adjacent submodules had best performance. distributing features correlation output submodule. Given current study, it expected will used general problems efficiently