Nuevas técnicas de particionamiento en clasificación automática

Abstract

In this article there are exposed some new thechniques for hte search of global optima inthe partitioning problem in Cluster Analysis. With these thechniques the results are sensibly improved with respect to the traditional methods. The methods developed here are well known in Combinatorial Optimization: i)simulated annealing; ii) tabu search, iii)genetic algorithms. We use these three approaches in the partitioning problem for clustering, following a search xhee similar to that of Regnier's algorithm of transfers. Keywords: Clustering, stochastic optimization, combinatorial optimization, heuristics.