Showing papers by "Enrique Alba published in 2006"

jMetal: a Java Framework for Developing Multi-Objective Optimization Metaheuristics

Abstract: This paper introduces jMetal, an object-oriented Java-based framework aimed at facilitating the development of metaheuristics for solving multi-objective optimization problems (MOPs). jMetal provides a rich set of classes which can be used as the building blocks of multi-objective metaheuristics; thus, taking advantage of code-reusing, the algorithms share the same base components, such as implementations of genetic operators and density estimators, so making the fair comparison of different metaheuristics for MOPs possible. The framework also incorporates a significant set of problems used as a benchmark in many comparative studies. We have implemented a number of multi-objective metaheuristics using jMetal, and an evaluation of these techniques using our framework is presented.

Efficient parallel LAN/WAN algorithms for optimization. The mallba project

Abstract: The MALLBA project tackles the resolution of combinatorial optimization problems using generic algorithmic skeletons implemented in C++. A skeleton in the MALLBA library implements an optimization method in one of the three families of generic optimization techniques offered: exact, heuristic and hybrid. Moreover, for each of those methods, MALLBA provides three different implementations: sequential, parallel for Local Area Networks, and parallel for Wide Area Networks. This paper introduces the architecture of the MALLBA library, details some of the implemented skeletons, and offers computational results for some classical optimization problems to show the viability of our library. Among other conclusions, we claim that the design used to develop the optimization techniques included in the library is generic and efficient at the same time.

Metaheuristic Procedures for Training Neutral Networks

Abstract: Classical Training Methods.- Local Search Based Methods.- Simulated Annealing.- Tabu Search.- Variable Neighbourhood Search.- Population Based Methods.- Estimation of Distribution Algorithms.- Genetic Algorithms.- Scatter Search.- Other Advanced Methods.- Ant Colony Optimization.- Cooperative Coevolutionary Methods.- Greedy Randomized Adaptive Search Procedures.- Memetic Algorithms.

Evaluation of parallel metaheuristics

Abstract: When evaluating algorithms a very important goal is to perform better than the state-of-the-art techniques.. This requires experimental tests to compare the new algorithm with respect to the rest. It is, in general, hard to make fair comparisons between algorithms such as metaheuristics. The reason is that we can infer di erent conclusions from the same results depending on the metrics we use and how they are applied. This is specially important for non-deterministic methods. This analysis becomes more complex if the study includes parallel metaheuristics, since many researchers are not aware of existing parallel metrics and their meanings, especially concerning the vast literature on parallel programming used well before metaheuristics were rst introduced. In this paper, we focus on the evaluation of parallel algorithms. We give a clear de nition of the main parallel performance metrics and we illustrate how they can be used.

Software testing with evolutionary strategies

Abstract: This paper applies the Evolutionary Strategy (ES) meta-heuristic to the automatic test data generation problem. The problem consists in creating automatically a set of input data to test a program. This is a required step in software development and a time consuming task in all software companies. We describe our proposal and study the influence of some parameters of the algorithm in the results. We use a benchmark of eleven programs that includes fundamental algorithms in computer science. Finally, we compare our ES with a Genetic Algorithm (GA), a well-known algorithm in this domain. The results show that the ES obtains in general better results than the GA for the benchmark used.

A Simple Cellular Genetic Algorithm for Continuous Optimization

Abstract: Cellular genetic algorithms (cGAs) are a kind of genetic algorithm (GA) – population based heuristic– with a structured population so that individuals can only interact with their neighbors. The existence of small overlapped neighborhoods in this decentralized population provides both diversity and exploration, while the exploitation of the search space is strengthened inside each neighborhood. This balance between exploration and exploitation makes cGAs naturally suitable for solving complex problems. In this paper we tackle the minimization of a number of problems (both academic and from the real world) with a real-coded cGA, called JCell. The results show that JCell improves the compared algorithms for a number of the studied problems, thus increasing the overall performance with respect to other complex heterogeneous distributed GAs, belonging to the state-of-the-art in continuous optimization.

Observations in using Grid-enabled technologies for solving multi-objective optimization problems

Abstract: In this paper, we analyze some technical issues concerning the use of Grid-enabled technologies based on the Globus Toolkit to solve multi-objective optimization problems. We develop two distributed algorithms: an enumerative search and an evolutionary algorithm. The former is a simple technique, whose high time complexity is mastered down to acceptable execution times to some extent by the use of a Grid-enabled computing system such Globus. The second algorithm is an extension of PAES, a sequential evolutionary algorithm. The parallel results indicate that using Globus is a promising research line to solve multi-objective problems in Grid computing environments.

Hierarchical cellular genetic algorithm

Abstract: Cellular Genetic Algorithms (cGA) are spatially distributed Genetic Algorithms that, because of their high level of diversity, are superior to regular GAs on several optimization functions. Also, since these distributed algorithms only require communication between few closely arranged individuals, they are very suitable for a parallel implementation. We propose a new kind of cGA, called hierarchical cGA (H-cGA), where the population structure is augmented with a hierarchy according to the current fitness of the individuals. Better individuals are moved towards the center of the grid, so that high quality solutions are exploited quickly, while at the same time new solutions are provided by individuals at the outside that keep exploring the search space. This algorithmic variant is expected to increase the convergence speed of the cGA algorithm and maintain the diversity given by the distributed layout. We examine the effect of the introduced hierarchy by observing the variable takeover rates at different hierarchy levels and we compare the H-cGA to the cGA algorithm on a set of benchmark problems and show that the new approach performs promising.

Comparative analysis of modern optimization tools for the p-median problem

Abstract: This paper develops a study on different modern optimization techniques to solve the p-median problem. We analyze the behavior of a class of evolutionary algorithm (EA) known as cellular EA (cEA), and compare it against a tailored neural network model and against a canonical genetic algorithm for optimization of the p-median problem. We also compare against existing approaches including variable neighborhood search and parallel scatter search, and show their relative performances on a large set of problem instances. Our conclusions state the advantages of using a cEA: wide applicability, low implementation effort and high accuracy. In addition, the neural network model shows up as being the more accurate tool at the price of a narrow applicability and larger customization effort.

Simulating Realistic Mobility Models for Large Heterogeneous MANETs

Abstract: Mobile ad hoc networks (MANETs) are composed of communicating mobile devices capable of spontaneously interconnecting without any pre-existing infrastructure. The wide spread of mobile devices (i.e. phones, PDAs, laptops) enables the deployment of metropolitan ad hoc networks, referred to as MobileMANs. Until recently, MobileMAN simulation suffered from a lack of appropriate tools. Therefore a new class of simulators dedicated to MobileMANs is appearing. This paper presents Mad hoc, a MANETs simulator which belongs to this class. In addition to providing particular models for the simulation of numerous nodes evolving in a metropolitan environment, Mad hoc comes with appropriate tools for the development and the monitoring of ad hoc applications. Mad hoc’s applications are presented.

Optimal placement of antennae using metaheuristics

Abstract: In this article we solve the radio network design problem (RND). This NP-hard combinatorial problem consist of determining a set of locations for placing radio antennae in a geographical area in order to offer high radio coverage using the smallest number of antennae. This problem is originally found in mobile telecommunications (such as mobile telephony), and is also relevant in the rising area of sensor networks. In this work we propose an evolutionary algorithm called CHC as the state of the art technique for solving RND problems and determine its expected performance for different instances of the RND problem.

Parallel Evolutionary Multiobjective Optimization

Abstract: Research on multiobjective optimization is very active currently because most of the real-world engineering optimization problems are multiobjective in nature. Multiobjective optimization does not restrict to find a unique single solution, but a set of solutions collectively known as the Pareto front. Evolutionary algorithms (EAs) are especially well-suited for solving such kind of problems because they are able to find multiple trade-off solutions in a single run. However, these algorithms may be computationally expensive because (1) real-world problem optimization typically involves tasks demanding high computational resources and (2) they are aimed at finding the whole front of optimal solutions instead of searching for a single optimum. Parallelizing EAs arises as a possible way of facing this drawback. The first goal of this chapter is to provide the reader with a wide overview of the literature on parallel EAs for multiobjective optimization. Later, we include an experimental study where we develop and analyze pPAES, a parallel EA for multiobjective optimization based on the Pareto Archived Evolution Strategy (PAES). The obtained results show that pPAES is a promising option for solving multiobjective optimization problems.

A Parallel Island Model for Estimation of Distribution Algorithms

Abstract: In this work we address the parallelization of the kind of Evolutionary Algorithms (EAs) known as Estimation of Distribution Algorithms (EDAs). After an initial discussion on the types of potentially parallel schemes for EDAs, we proceed to design a distributed island version (dEDA), aimed at improving the numerical efficiency of the sequential algorithm in terms of the number of evaluations. After evaluating such a dEDA on several well-known discrete and continuous test problems, we conclude that our model clearly outperforms existing centralized approaches from a numerical point of view, as well as speeding up the search considerably, thanks to its suitability for physical parallelism.

Evolutionary algorithms in telecommunications

Abstract: Telecommunications are an important symbol of our present information society. With a rapidly growing number of user services, telecommunications is a field in which many open research lines are challenging the research community. Many of the problems found in this area can be formulated as optimization tasks. Some examples are assigning frequencies in radio link communications, developing error correcting codes for transmission of messages, and designing the telecommunication network. In practice, most of these optimization tasks are unaffordable with exact techniques. Hence, the utilization of heuristic approaches is in order. In this sense, evolutionary algorithms have constituted a popular choice. This paper summarizes some applications of the telecommunication field in which evolutionary algorithms have been successfully applied.

Theory and practice of cellular UMDA for discrete optimization

Abstract: A new class of estimation of distribution algorithms (EDAs), known as cellular EDAs (cEDAs), has recently emerged. In these algorithms, the population is decentralized by partitioning it into many small collaborating subpopulations, arranged in a toroidal grid, and interacting only with its neighboring subpopulations. In this work, we study the simplest cEDA —the cellular univariate marginal distribution algorithm (cUMDA). In an attempt to explain its behaviour, we extend the well known takeover time analysis usually applied to other evolutionary algorithms to the field of EDAs. We also give in this work empirical arguments in favor of using the cUMDAs instead of its centralized equivalent.

Optimal broadcasting in metropolitan MANETs using multiobjective scatter search

Abstract: Mobile Ad-hoc Networks (MANETs) are composed of a set of communicating devices which are able to spontaneously interconnect without any pre-existing infrastructure. In such scenario, broadcasting becomes an operation of capital importance for the own existence and operation of the network. Optimizing a broadcasting strategy in MANETs is a multiobjective problem accounting for three goals: reaching as many stations as possible, minimizing the network utilization, and reducing the makespan. In this paper, we face this multiobjective problem with a state-of-the-art multiobjective scatter search algorithm called AbSS (Archive-based Scatter Search) that computes a Pareto front of solutions to empower a human designer with the ability of choosing the preferred configuration for the network. Results are compared against those obtained with the previous proposal used for solving the problem, a cellular multiobjective genetic algorithm (cMOGA). We conclude that AbSS outperforms cMOGA with respect to three different metrics.

Workforce planning with parallel algorithms

Abstract: Workforce planning is an important activity that enables organizations to determine the workforce needed for continued success. A workforce planning problem is a very complex task that requires modern techniques to be solved adequately. In this work, we describe the development of two parallel metaheuristic methods, a parallel genetic algorithm and a parallel scatter search, which can find high-quality solutions to 20 different problem instances. Our experiments show that parallel versions do not only allow to reduce the execution time but they also improve the solution quality.

ROS (Remote Optimization Service)

Abstract: Resumen. En este informe se describe el Servicio de Optimizacion Remota ROS. Este servicio permite a sus clientes el acceso remoto con la posibilidad de ejecutar y manipular algoritmos de optimizacion basados en tecnicas heuristicas, situados en repositorios con recursos hardware y software para optimizar problemas practicos y reales. Presentamos tambien la relacion con otros trabajos similares, su arquitectura y principales caracteristicas. Ademas, se realiza una introduccion de las herramientas de comunicacion utilizadas, especi cacion de los tipos de datos con XML y encapsulado de algoritmos a traves del servicio ROS. Por ultimo, se muestra una comparativa sobre las distintas formas de con gurar este sistema, asi como de los resultados obtenidos tras realizar evaluaciones de este servicio sobre distintas con guraciones tipicas.

Avancées des métaheuristiques pour l'optimisation combinatoire multi-objectif

Abstract: Ce document presente certaines voies prometteuses, emergent actuellement dans le domaine de l'optimisation combinatoire multiobjectif Resoudre de tels problemes implique notamment la recherche d'un ensemble de solutions dites ``Pareto optimales'' Ces solutions sont les meilleurs compromis realisable pour les differents objectifs a optimiser pour le probleme etudie, le but etant de decouvrir un ensemble de bonne qualite en terme de convergence, mais egalement en terme de diversite des compromis proposes Dans le domaine des metaheuristiques, il existe plusieurs etat de l'art du domaine traitant principalement des algorithmes evolutionnaires Nous nous proposons ici d'enrichir ces etudes en relevant des approches recentes qui ont fait preuve d'innovation mais egalement de bons resultats Apres une introduction generale et avoir propose une classification des methodes usuelles, nous nous proposons de discuter des orientations recentes et prometteuses de la recherche dans ce domaine Les approches etudiees sont l'application des metaheuristues mono-objectif recentes au cadre multi-objectif, les metaheuristiques hybrides, les metaheuristiques multi-objectif et le parallelisme, et enfin l'optimisation multi-objectif sous incertitude Nous concluerons par une discussion et quelques questions ouvertes

The Bayesian method of moments (BMOM) in some aggregation problems in econometrics

Abstract: The BMOM is particularly useful for obtaining post-data moments and densities for parameters and future observations when the form of the likelihood function is unknown and thus a traditional Bayesian approach cannot be used. Also, even when the form of the likelihood is assumed known, in time series problems it is sometimes difficult to formulate an appropriate prior density. Here, we show how the BMOM approach can be used in two, nontraditional problems. The first one is conditional forecasting in regression and time series autoregressive models. Specifically, it is shown that when forecasting disaggregated data (say quarterly data) and given aggregate constraints (say in terms of annual data) it is possible to apply a Bayesian approach to derive conditional forecasts in the multiple regression model. The types of constraints (conditioning) usually considered are that the sum, or the average, of the forecasts equals a given value. This kind of condition can be applied to forecasting quarterly values whose sum must be equal to a given annual value. Analogous results are obtained for AR(p) models. The second problem we analyse is the issue of aggregation and disaggregation of data in relation to predictive precision and modelling. Predictive densities are derived for future aggregate values by means of the BMOM based on a model for disaggregated data. They are then compared with those derived based on aggregated data. Copyright © 2006 John Wiley & Sons, Ltd.

Solving the illumination problem with heuristics

Abstract: In this article we propose optimal and quasi optimal solutions to the problem of searching for the maximum lighting point inside a polygon P of n vertices. This problem is solved by using three different techniques: random search, simulated annealing and gradient. Our comparative study shows that simulated annealing is very competitive in this application. To accomplish the study, a new polygon generator has been implemented, which greatly helps in the general validation of our claims on the illumination problem as a new class of optimization task.