Ensemble-based Multi-objective Clustering Algorithms for Gene Expression Data Sets

In this paper, two multi-objective clustering ensemble algorithms are proposed named MOCLED and MOCNCD. MOCLED is different from MOCLE on three points. First, different clustering algorithms are used to produce some new individuals in evolutionary process. Second, a new screening mechanism is added. In each generation, the worst individual is replaced by the best individual. Third, a new objective function is added to ensure a diverse population. MOCNCD is the same as MOCLED except the crossover operator. We replace it with a new proposed cluster ensemble algorithm, IDICLENS. Experimental results reveal the advantages of our method on finding good partitions.