What Makes a Problem Hard for a Genetic Algorithm? Some Anomalous Results and Their Explanation
Machine Learning - Special issue on genetic algorithms
Evolutionary algorithms in theory and practice: evolution strategies, evolutionary programming, genetic algorithms
An introduction to genetic algorithms
An introduction to genetic algorithms
A Methodology for the Statistical Characterization of Genetic Algorithms
MICAI '02 Proceedings of the Second Mexican International Conference on Artificial Intelligence: Advances in Artificial Intelligence
Locally adaptive metrics for clustering high dimensional data
Data Mining and Knowledge Discovery
A robust evolutionary algorithm for the recovery of rational Gielis curves
Pattern Recognition
Hi-index | 0.00 |
Clustering is the task which allows us to identify groups, distributions or patterns over a set of data. To achieve it we must, first, adopt a measure of likelihood which allows us to group elements of similar characteristics into two or more such clusters. Presently there are several clustering algorithms which appeal to various metrics among which there outstand the ones based on Minkowski's and Mahalanobis' distances. In general this approach yields hyperspherical forms which restrict the fact that some clusters may be represented by irregular N-dimensional bodies. In this paper we use a formula originally developed by Johan Gielis which has been called the "superformula". It allows the generation of N-dimensional bodies of arbitrary shape by modifying certain parameters. This approach allows us to represent a cluster as the set of data contained in a given body without resorting to a metric of distance. Therefore, we replace the idea of nearness by one of membership. To determine the more adequate values of the parameters in the superformula we applied a Genetic Algorithm (GA); the so called Vasconcelos" GA.