An ensemble game theoretic approach for multi-objective optimization

Article type: Research Article

Authors: Badami, Mahsa | Mozafari, Niloofar | Hamzeh, Ali^{; *} | Hashemi, Sattar

Affiliations: Department of Computer Science, Engineering and IT, School of Electrical and Computer Engineering, Shiraz University, Iran. E-mails: [email protected], [email protected], [email protected], [email protected]

Correspondence: [*] Corresponding author. E-mail: [email protected].

Abstract: Recently multi-objective clustering has been extensively explored due to its appearance of new applications in many domains. However, in many applications, there is more than a single objective which is needed to be optimized in the context of the application, such as facility location, ad hoc networks and sensor networks. These domains must optimize two objectives of compactness and equi-partitioning which may be conflicted in some situations. Existing algorithms have high complexity. In this paper, we propose an Ensemble Game Theoretic approach for multi-objective clustering method which optimizes two objectives of compactness and equi-partitioning, simultaneously. We compare our algorithm on variety of data sets including synthetic and real ones. The remarkable results are very promising and demonstrate the efficiency of presented approach both in performance and complexity.

Keywords: Multi-objective clustering, game theory, ensemble clustering, equi-partitioning, compactness

DOI: 10.3233/AIC-140653

Journal: AI Communications, vol. 28, no. 3, pp. 553-566, 2015