Using gravitational search algorithm in prototype generation for nearest neighbor classification

• Published in: Neurocomputing (Amsterdam) Vol. 157; pp. 256 - 263
• Main Authors: Rezaei, Mohadese, Nezamabadi-pour, Hossein
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01-06-2015
• Subjects: Classification; Gravitational search algorithm; K-nearest neighbor; Prototype generation; Gravitational search algorithm; Prototype generation; Classification; K-nearest neighbor
• ISSN: 0925-2312; 1872-8286
• DOI: 10.1016/j.neucom.2015.01.008

In recent years, metaheuristic algorithms have emerged as a promising approach to solve clustering and classification problems. In this paper, gravitational search algorithm (GSA) which is one of the newest swarm based metaheuristic search techniques, is adapted to generate prototypes for nearest neighbor classification. The proposed method has been tested on several problems and the results are compared with those obtained by several state-of-the-art techniques. The comparison shows that our proposed method can achieve higher classification accuracy than the competing methods and has good performance in the field of prototype generation.