An Improved Deep Convolutional Fuzzy System for Classification Problems
Huidong Wang, Jinli Yao · 2020
Fuzzy rule-based classifier is an effective classification algorithm. However, traditional fuzzy system is faced with the challenge of rule explosion and low training speed when dealing with big data problems. In this paper, an improved deep convolutional fuzzy system (DCFS) is proposed for big data classification problems. First, an improved Wang- Mendel (WM) Method is put forward for the training of each sub-fuzzy system. Second, a hierarchical DCFS is constructed for big data classification problems. The system performance is demonstrated via simulation experiments on a couple of classification datasets of varying sizes.