Interpretable classifier design by axiomatic fuzzy sets theory and derivative-free optimization

Yuangang Wang, Jiaming Duan, Haoran Liu, Shuo Guan, Xiaodong Liu, Xiaodong Duan

Abstract: Highlights•Develop encoding–decoding mechanism to bridge complex concept and solution vector.•Design various fitness function to balance prediction accuracy and interpretability.•Covert training AFS-based classifier to solving combinatorial optimization problem.•Generate semantic description for prediction result in form of natural language.