This paper introduces a cultural algorithm that uses domain knowledge to improve the performance of an evolutionary programming technique adopted for constrained optimization. The proposed approach extracts during process and builds map feasible region guide search more efficiently. Additionally, in order have efficient memory management scheme, current implementation 2 n -trees store this region. Results indicate is able produce very competitive results with respect other optimization...

This article provides a short introduction to the evolutionary multiobjective optimization field. The first part of discusses most representative algorithms that have been developed, from historical perspective. In second article, some applications within materials science and engineering are reviewed. final potential areas for future research in this area briefly described.

In this paper, we present the first proposal to use a cultural algorithm solve multiobjective optimization problems. Our uses evolutionary programming, Pareto ranking and elitism (i.e., an external population). The approach proposed is validated using several examples taken from specialized literature. results are compared with respect NSGA-II, which representative of state-of-the-art in optimization. performance our indicates that algorithms viable alternative for

This paper presents a simple (1 + /spl lambda/) evolution strategy and three selection criteria to solve engineering optimization problems. approach avoids the use of penalty function deal with constraints. Its main advantage is that it does not require definition extra parameters, other than those used by strategy. A self-adaptation mechanism allows algorithm maintain diversity during process in order reach competitive solutions at low computational cost. The was tested four well-known...

In recent years, the development of selection mechanisms based on performance indicators has become an important trend in algorithmic design. Hereof, hypervolume been most popular choice. Multi-objective evolutionary algorithms (MOEAs) this indicator seem to be a good choice for dealing with many-objective optimization problems. However, their main drawback is that such are typically computationally expensive. This motivated some research which use other explored. Here, we propose efficient...

In this paper we propose a cultural algorithm, where different knowledge sources modify the variation operator of differential evolution algorithm. Differential is used as basis for population, and selection processes. The experiments performed show that cultured able to reduce number fitness function evaluations needed obtain good aproximation optimum value in constrained real-parameter optimization. Comparisons are provided with respect three techniques representative state-of-the-art area.

We propose the use of differential evolution as a population space cultural algorithm, applied to solution constrained optimization problems. Differential is relatively recent evolutionary algorithm that has been found be very robust search engine for real parameter optimization. Adding different knowledge sources variation operator it possible improve and reduce computational cost necessary approximate global optima The proposed technique validated using set well-known problems commonly...

One of the main tasks software testing involves is generation test inputs to be used during test. Due its expensive cost, automation this task has become one key issues in area. Recently, been explicitly formulated as resolution a set constrained optimisation problems. Differential Evolution (DE) population based evolutionary algorithm which successfully applied number domains, including optimisation. We present data generator employing DE solve each problems, and empirically evaluate...

This article presents a novel method to compute averaged Hausdorff () approximations of the Pareto fronts multi-objective optimization problems. The underlying idea is utilize directly scalar problem that induced by performance indicator. can be viewed as certain set based scalarization approach and addressed both mathematical programming techniques evolutionary algorithms (EAs). In this work, focus on latter where first single objective EA for such proposed. Finally, strength demonstrated...

In this work, an approach for solving the job shop scheduling problem using a cultural algorithm is proposed. Cultural algorithms are evolutionary computation methods that extract domain knowledge during process. Additional to extracted knowledge, proposed also uses given priori (based on specific available problem). The compared with respect Greedy Randomized Adaptive Search Procedure (GRASP), Parallel GRASP, Genetic Algorithm, Hybrid and deterministic method called shifting bottleneck. in...

This work proposes the use of a micro genetic algorithm to optimize architecture fully connected layers in convolutional neural networks, with aim reducing model complexity without sacrificing performance. Our approach applies paradigm transfer learning, enabling training need for extensive datasets. A requires fewer computational resources due its reduced population size, while still preserving substantial degree search capabilities found algorithms larger populations. By exploring...

In this paper, we propose the combination of different optimization techniques in order to solve "hard" two- and three-objective problems at a relatively low computational cost. First, use ε-constraint method obtain few points over (or very near of) true Pareto front, then an approach based on rough sets spread these solutions, so that entire front can be covered. The constrained single-objective optimizer required by method, is cultured differential evolution, which efficient for...

This paper presents the goal-constraint method for incorporating preferences in multiobjective optimization. The are provided form of a vector goals, which is familiar decision makers and operations researchers. portion Pareto front to be generated totally defined by regardless if such feasible or not. Once defined, it experiment on many objective problems, because reduced cost producing less points. experimental results show good convergence properties, graphs illustrate way produced related goals.