Quantum-inspired evolutionary multi-objective fuzzy classifier with real and categorical representation

Article type: Research Article

Authors: Nunes, Waldir | Vellasco, Marley^{; *} | Tanscheit, Ricardo

Affiliations: Department of Electrical Engineering, Pontifical Catholic University of Rio de Janeiro, Brazil

Correspondence: [*] Corresponding author. Marley Vellasco, Department of Electrical Engineering, Pontifical Catholic University of Rio de Janeiro, Brazil. E-mail: [email protected].

Abstract: This work presents a new model for the automatic synthesis of fuzzy classifiers, based on quantum-inspired evolutionary algorithms, which overcomes the difficulties inherent to the use of hybrid representations and the treatment of multiple objectives, both necessary for the synthesis of these types of systems. Without any a priori information about the classifier rules or any initial adjustment of individuals, the results obtained are comparable to those of other techniques that start from classifier populations previously adjusted to obtain good performance. According to the current trend, the aim was to build classifiers with good accuracy and simultaneous high interpretability of their fuzzy rule base.

Keywords: Fuzzy classifier, quantum-inspired evolutionary algorithms, multi-objective optimization

DOI: 10.3233/JIFS-181710

Journal: Journal of Intelligent & Fuzzy Systems, vol. 36, no. 6, pp. 5875-5887, 2019