An exact algorithm for fuzzy discriminant analysis

Aiming to minimize the sum of classification errors, Lin and Chen recently proposed a method that uses a genetic algorithm to perform fuzzy discriminant analysis. The proposed genetic algorithm efficiently obtains near optimal solutions but suffers the problem of providing varied and non-optimal solutions to a given problem because of using the stochastic genetic algorithm. This study devises an exact algorithm for fuzzy discriminant analysis, which uses a mixed 0-1 programming model to minimize the sum of the absolute or squared classification errors. Numerical examples demonstrate the effectiveness of the proposed method. The proposed method is easy to implement and provides optimal solutions to fuzzy discriminant analysis problems.