Clustering methods have proved very successful in tackling the global optimization problem. One reason is that they make it possible to very efficiently combine global and local search.
The clustering technique facilitate efficient representation of the information on the problem encountered in the solution process. The output from the process showing the evolution of clusters contains a lot of extra information hard to formalize, but of great importance for the practitioner solving a real world problem.
In clustering methods stopping conditions available for uniform sampling and multistart relating the goal prescribed to the effort needed to achieve this goal can be used. Considerable theoretical progress has been made in this direction. The clustering approach to global optimization thus rests on a sound theoretical base which makes these methods attractive both from a practical and mathematical standpoint.
KeywordsLocal Minimum Cluster Algorithm Local Search Global Optimization Cluster Method
Unable to display preview. Download preview PDF.