Catálogo de publicaciones - libros

The reliance of Tabu Search (TS) algorithms on a local search leads to a logical development of algorithms that use more than one search concurrently. In this paper we present a multi-threaded TS algorithm employing a number of threads that share information. We assess the performance of this algorithm compared to previous multi-objective TS algorithms, via the results obtained from applying the algorithms to a range of standard test functions. We also consider whether an optimal number of threads can be found, and what impact changing the number of threads used has on performance. We discover that, contrary to the popular belief that multi-threading is usually beneficial, performance only improves in a few special cases.

This paper presents a comprehensive comparison between the performance of state-of-the-art genetic algorithms NSGA-II, SPEA2 and IBEA and their differential evolution based variants DEMO, DEMO and DEMO. Experimental results on 16 numerical multiobjective test problems show that on the majority of problems, the algorithms based on differential evolution perform significantly better than the corresponding genetic algorithms with regard to applied quality indicators. This suggests that in numerical multiobjective optimization, differential evolution explores the decision space more efficiently than genetic algorithms.

This paper presents the Efficient Multi-Objective Particle Swarm Optimizer (EMOPSO), which is an improved version of a multi-objective evolutionary algorithm (MOEA) previously proposed by the authors. Throughout the paper, we provide several details of the design process that led us to EMOPSO. The main issues discussed are: the mechanism to maintain a set of well-distributed nondominated solutions, the turbulence operator that avoids premature convergence, the constraint-handling scheme, and the study of parameters that led us to propose a self-adaptation mechanism. The final algorithm is able to produce reasonably good approximations of the Pareto front of problems with up to 30 decision variables, while performing only 2,000 fitness function evaluations. As far as we know, this is the lowest number of evaluations reported so far for any multi-objective particle swarm optimizer. Our results are compared with respect to the NSGA-II in 12 test functions taken from the specialized literature.

To find solutions as close to the Pareto front as possible, and to make them as diverse as possible in the obtained non-dominated front is a challenging task for any multiobjective optimization algorithm.-dominance is a concept which can make genetic algorithm obtain a good distribution of Pareto-optimal solutions and has low computational time complexity,and the orthogonal design method can generate an initial population of points that are scattered uniformly over the feasible solution space.In this paper, combining -dominance and orthogonal design method, we propose a novel Differential Evolution (DE) algorithm for multiobjective optimization .Experiments on a number of two- and three-objective test problems of diverse complexities show that our approach is able to obtain a good distribution with a small computational time in all cases. Compared with several other state-of-the-art evolutionary algorithms, it achieves not only comparable results in terms of convergence and diversity metrics, but also a considerable reduction of the computational effort.

Molecular system possesses two main characteristics that seem to be applicable for the contrary goals of proximity and diversity in multiobjective optimization, namely the converging pressure in potential fields as dictated by the Maxwell-Boltzmann distribution and the inherent drift to a homogenous and uniform equilibrium with maximum entropy, even without any prior knowledge on the geometry and state of the enclosure. Inspired by this association, this paper explores the notion of exploiting molecular motion to solve multiobjective problems. By adapting the algorithmic structure of molecular dynamics, which essentially represents a technique for the computer simulation of molecular motion, a molecular system that is relevant for multiobjective optimization is proposed, known as molecular dynamics optimizer (MDO). The performance of MDO was subsequently compared with other conventional multiobjective optimizers, specifically EA and PSO, and the experimental results demonstrated that MDO is indeed a viable and practical approach for multiobjective optimization.

One of main issues in multi-objective optimization is to support for choosing a final solution from Pareto frontier which is the set of solution to problem. For generating a part of Pareto optimal solution closest to an aspiration level of decision maker, not the whole set of Pareto optimal solutions, we propose a method which is composed of two steps; i) approximate the form of each objective function by using support vector regression on the basis of some sample data, and ii) generate Pareto frontier to the approximated objective functions based on given the aspiration level. In addition, we suggest to select additional data for approximating sequentially the forms of objective functions by relearning step by step. Finally, the effectiveness of the proposed method will be shown through some numerical examples.

The design of a Hybrid Electric Vehicle (HEV) system is an energy management strategy problem between two sources of power. Traditionally, the drive train has been designed first, and then a driving strategy chosen and sometimes optimised. This paper considers the simultaneous optimisation of both drive train and driving strategy variables of the HEV system through use of a multi-objective evolutionary optimiser. The drive train is well understood. However, the optimal driving strategy to determine efficient and opportune use of each prime mover is subject to the driving cycle (the type of dynamic environment, e.g. urban, highway), and has been shown to depend on the correct selection of the drive train parameters (gear ratios) as well as driving strategy heuristic parameters. In this paper, it is proposed that the overall optimal design problem has to consider multiple objectives, such as fuel consumption, reduction in electrical energy stored, and the ‘driveability’ of the vehicle. Numerical results shows improvement when considering multiple objectives and simultaneous optimisation of both drive train and driving strategy.

This paper will investigate the application of multiobjective evolu-tionary neural networks in time series forecasting. The proposed algorithmic model considers training and validation accuracy as the objectives to be optimized simultaneously, so as to balance the accuracy and generalization of the evolved neural networks. To improve the overall generalization ability for the set of solutions attained by the multiobjective evolutionary optimizer, a simple algorithm to filter possible outliers, which tend to deteriorate the overall performance, is proposed also. Performance comparison with other existing evolutionary neural networks in several time series problems demonstrates the practicality and viability of the proposed time series forecasting model.

Understanding the results of a multi objective optimization process can be hard. Various visualization methods have been proposed previously, but the only consistently popular one is the 2D or 3D objective scatterplot, which cannot be extended to handle more than 3 objectives. Additionally, the visualization of high dimensional parameter spaces has traditionally been neglected. We propose a new method, based on heatmaps, for the simultaneous visualization of objective and parameter spaces. We demonstrate its application on a simple 3D test function and also apply heatmaps to the analysis of real-world optimization problems. Finally we use the technique to compare the performance of two different multi-objective algorithms.

Multiplex Polymerase Chain Reaction (PCR) assay is to amplify multiple target DNAs simultaneously using different primer pairs for each target DNA. Recently, it is widely used for various biology applications such as genotyping. For sucessful experiments, both the primer pairs for each target DNA and grouping of targets to be actually amplified in one tube should be optimized. This involves multiple conflicting objectives such as minimizing the interaction of primers in a group and minimizing the number of groups required for the assay. Therefore, a multiobjective evolutionary approach may be an appropriate approach. In this paper, a hybrid multiobjective evolutionary algorithm which combines -multiobjective evolutionary algorithm with local search is proposed for multiplex PCR assay design. The proposed approach was compared with another multiobjective method, called MuPlex, and showed comparative performance by covering all of the given target sequences.