Evolutionary Algorithms for Multiobjective Optimization: Methods and Applications


Abstract

Many real-world problems involve two types of problem difficulty: i) multiple, conflicting objectives and ii) a highly complex search space. On the one hand, instead of a single optimal solution competing goals give rise to a set of compromise solutions, generally denoted as Pareto-optimal. In the absence of preference information, none of the corresponding trade-offs can be said to be better than the others. On the other hand, the search space can be too large and too complex to be solved by exact methods. Thus, efficient optimization strategies are required that are able to deal with both difficulties. Evolutionary algorithms possess several characteristics that are desirable for this kind of problem and make them preferable to classical optimization methods. In fact, various evolutionary approaches to multiobjective optimization have been proposed since 1985, capable of searching for multiple Pareto-optimal solutions concurrently in a single simulation run. However, in spite of this variety, there is a lack of extensive comparative studies in the literature. Therefore, it has remained open up to now: i) whether some techniques are in general superior to others, ii) which algorithms are suited to which kind of problem, and iii) what the specific advantages and drawbacks of certain methods are. The subject of this work is the comparison and the improvement of existing multiobjective evolutionary algorithms and their application to system design problems in computer engineering. In detail, the major contributions are: - An experimental methodology to compare multiobjective optimizers is developed. In particular, quantitative measures to assess the quality of trade-off fronts are introduced and a set of general test problems is defined, which are i) easy to formulate, ii) represent essential aspects of real-world problems, and iii) test for different types of problem difficulty. - On the basis of this methodology, an extensive comparison of numerous evolutionary techniques is performed in which further aspects such as the influence of elitism and the population size are also investigated. - A novel approach to multiobjective optimization, the strength Pareto evolutionary algorithm, is proposed. It combines both established and new techniques in a unique manner. - Two complex multicriteria applications are addressed using evolutionary algorithms: i) the automatic synthesis of heterogeneous hardware/systems and ii) the multidimensional exploration of software implementations for digital signal processors.