Merged Clustering Imperialists Algorithm (MCIA)
Mohammad Amin Soltani-Sarvestani · International Journal of Information and Electronics Engineering · 2012
This paper introduces an improved evolutionary algorithm based on the imperialist competitive algorithm (ICA) called merged clustering imperialists algorithm (MCIA). Merged clustering imperialist algorithm is another version of clustering imperialist algorithm (CIA). The imperialist competitive algorithm that has recently been introduced is used in order to optimize problems and has shown its reliable performance. This novel optimization algorithm is inspired by socio-political process of imperialistic competition in the real world. In the ICA, there are two categories of countries: Imperialists and Colonies that each Imperialist observes its colonies by absorption policy. In the proposed algorithm, the changed ICA is used for clustering data. In the MCIA, according to the number of considered clusters, the Imperialists will be created, and they act like Imperialists in the ICA and absorb their colonies. In this process which colonies move toward Imperialists, the Imperialists borders and regions are specified and by this way whole region is clustered into the number of Imperialists. When the region and border of clusters are specified, with a Merge operator that is different in various clustering problems, the clusters' components merge to each other. Finally, the proposed algorithm is used for clustering an image and a desired result is obtained. movement towards imperialist's position dynamically. In the imperialist competitive algorithm (ICA) there are two different types of countries, imperialists and colonies, which imperialists absorb their colonies. The imperialists use absorption policy to absorb colonies in different dimensions. In the real world, each country and each imperialist has a specific border which is usually clarified by some spots. In the proposed algorithm each imperialist is considered as the center of each cluster and absorbs colonies toward itself by absorption policy. When colonies move toward imperialists specify the border between them. In this approach the number of considered clusters, the imperialists will be created and colonies specify the borders spots between different clusters during their movement toward Imperialists. So a feature vector is considered and each colony by comparison with its feature vector with all imperialists' vectors recognizes that it is placed in which imperialist and clarifies the border between imperialists. The rest of this paper is arranged in this order: Section II describes the brief description of imperialist competitive algorithm, Section III presents the proposed algorithm. In Section IV, the result will be analyzed and the performance of algorithms will be evaluated. Section V draws a conclusion from this paper.