Grouping Genetic Algorithm for Data Clustering
Clustering can be visualized as a grouping problem as it consists of identifying finite set of groups in a dataset. Grouping genetic algorithms are specially designed to handle grouping problems. As the clustering criteria such as minimizing the with-in cluster distance is high-dimensional, non-linear and multi-modal, many standard algorithms available in the literature for clustering tend to converge to a locally optimal solution and/or have slow convergence. Even genetic guided clustering algorithms which are capable of identifying better quality solutions in general are also not totally immune to these shortcomings because of their ad hoc approach towards clustering invalidity and context insensitivity. To remove these shortcomings we have proposed a hybrid steady-state grouping genetic algorithm. Computational results show the effectiveness of our approach.
KeywordsGrouping Genetic Algorithm Data Clustering Heuristic
Unable to display preview. Download preview PDF.
- 4.Painho, M., Fernando, B.: Using genetic algorithms in clustering problems. In: Proceedings of the 5th International Conference on GeoComputation (2000)Google Scholar