Genetic Optimized Fuzzy Extreme Learning Machine Ensembles for Affect Classification
Wei Shiung Liew, Chu Kiong Loo, Takenori Obo · 2016
This paper presents a method for generating an optimized ensemble of fuzzy extreme learning machines (FELM) using a combination of genetic algorithms with a Bayesian Information Criterion (GA-BIC) fitness function. The operation of the FELM is equivalent to that of a fuzzy inference system, and is used for learning and classifying a given data set. The relative simplicity of the FELM structure enables a large number of FELMs to be generated in a short time. GA-BIC is used to select the minimum number of FELMs while maximizing the effectiveness of the classifier ensemble.