ESTIMATION OF DISTRIBUTION ALGORITHM FOR MULTIOBJECTIVE OPTIMIZATION OF HYBRID METAHEURISTICS
ESTIMATION OF DISTRIBUTION ALGORITHM FOR MULTIOBJECTIVE OPTIMIZATION OF HYBRID METAHEURISTICS

Summary/Abstract: In most real issues, whether technological, economic otherwise, the ability to choose the decisions to be made is essential. A decision maker needs to know the problem very well so that the choices can be optimal. However, real-world problems have several goals to achieve, so finding solutions that optimize goals simultaneously fall into the realm of multi-goal optimization. Solving multi objective optimization problems aims to find a set of solutions called Pareto. Pareto solutions are those that are not dominated by any other solution. Several methods have been used to solve this problem, of which evolutionary algorithms are the most appropriate. This is due to their ability to find good approximations of Pareto solutions. To deal with this problem, the algorithm used must be able to model the solutions obtained during the optimization phase. Such a model will be exploited if new solutions are needed. For this, a distribution estimation algorithm is used. It aims to estimate the distribution of the best solutions that will be used to generate new solutions with the same characteristics of optimism. The aim is to minimize execution time in the solution update phase, especially in multi-objective issues. In this paper, it is presented an algorithm proposed by Shi, Ma and Yang which models a promising area in the search space using a probability model, and this model is used to produce new solutions. It is specified that the variables used are those in the small space, this technique is used to reduce operating costs.

• Details
• Contents