A Multi-Objective Memetic Algorithm Based Clustering Method

Data clustering is a challenging problem where clustering algorithms are only based on one criterion, which is able to reveal all types of structures presenting in the data. It is unfeasible to establish a priori clustering criterion which is supposed to be more appropriate to capture the structure in the data. Clustering algorithms with different objective or criterion are essential to apply to the data in order to obtain different structures. This paper presents an attempt of using Multi-Objectives Memetic Algorithm (MOMA) for providing multi-objectives clustering. The approach is supposed to provide better clustering result with complex structured data set, as well as better ability to define optimal number of clustering with direct optimization on cluster validation.