Clustering with a genetically optimized approach

This paper describes a genetically guided approach to optimizing the hard (J1) and fuzzy (Jm) c-means functionals used in cluster analysis. Our experiments show that a genetic algorithm ameliorates the di culty of choosing an initialization for the c-means clustering algorithms. Experiments use six data sets, including the Iris data, magnetic resonance and color images. The genetic algorithm approach is generally able to nd the lowest known Jm value or a Jm associated with a partition very similar to that associated with the lowest Jm value. On data sets with several local extrema, the GA approach always avoids the less desirable solutions. Degenerate partitions are always avoided by the GA approach, which provides an e ective method for optimizing clustering models whose objective function can be represented in terms of cluster centers. The time cost of genetic guided clustering is shown to make a series random initializations of fuzzy/hard c-means, where the partition associated with the lowest Jm value is chosen, an e ective competitor for many clustering domains.