A Study on Two-Stage Multi-objective Fuzzy Genetics-based Machine Learning Using an Archive Population

Abstract

Multi-objective fuzzy genetics-based machine learning (MoFGBML), one of the most well-known multi-objective evolutionary fuzzy systems, can efficiently obtain a set of fuzzy classifiers considering the maximization of classification performance and the minimization of the model complexity. However, MoFGBML has a strong bias towards minimizing complexity in the search process, which makes it easy to obtain classifiers with low complexity but difficult to obtain classifiers with high classification performance. As a result, the number of non-dominated classifiers is often small. In our previous study, two-stage fuzzy genetics-based machine learning has been proposed to mitigate this bias: first, an accuracy-oriented single-objective optimization is performed, and then a multi-objective optimization is performed to maximize the classification performance and minimize the complexity. In this paper we consider the use of an archive population in two-stage fuzzy genetics-based machine learning to further increase the number of non-dominated solutions and improve the tradeoff curve between two objectives.