Case-base maintenance based on representative selection for 1-NN algorithm

Abstract

Case-based reasoning (CBR) uses known experiences to solve new problems. Past problems are stored as cases in a case base and a new case is classified by determining the most similar case from the case base. The nearest neighbor (NN) algorithm is one of the most basic CBR and case-base maintenance (CBM) is an important issue in CBR system to obtain the efficient case bases. This paper proposes a new approach to selection of the representative cases based on generalization capability of cases. Using this method, most redundant cases which affect the solution accuracy is deleted. The experiments show that the proposed method can remove greatly the redundant cases, as well as preserve a satisfying accuracy of solutions when it is used in 1-NN algorithm for classification tasks.