Stepwise Adaption of Weights with Refinement and Decay on Constraint Satisfaction Problems

Source: CiteSeer


Adaptive fitness functions have led to very successful evolutionary algorithms (EA) for various types of constraint satisfaction problems (CSPs). In this paper we consider one particular fitness function adaptation mechanism, the so called Stepwise Adaption of Weights (SAW). We compare algorithm variants including two penalty systems and we experiment with extensions of the SAW mechanism utilizing a refinement function and a decay function. Experiments are executed on binary CSP instances generated by a recently proposed method (method E).

Download full-text


Available from: Bart Craenen, Sep 18, 2012
1 Follower
14 Reads
  • Source
    • "The constraints can be handled directly, indirectly or by mixed methods [14]. Among the available methods, some put the emphasis on the usage of heuristics, such as ARC-GA [15], [16], COE-H GA [17], [18], Glass-Box [19], H-GA [20], [21], whereas others handle the constraints by fitness function adaptation, such as CCS [22], [23], MID [24]-[26], SAW [27], [28]. These methods dealt with CSPs by EAs from different angles and boosted the development of this field. "
    [Show abstract] [Hide abstract]
    ABSTRACT: With the intrinsic properties of constraint satisfaction problems (CSPs) in mind, we divide CSPs into two types, namely, permutation CSPs and nonpermutation CSPs. According to their characteristics, several behaviors are designed for agents by making use of the ability of agents to sense and act on the environment. These behaviors are controlled by means of evolution, so that the multiagent evolutionary algorithm for constraint satisfaction problems (MAEA-CSPs) results. To overcome the disadvantages of the general encoding methods, the minimum conflict encoding is also proposed. Theoretical analyzes show that MAEA-CSPs has a linear space complexity and converges to the global optimum. The first part of the experiments uses 250 benchmark binary CSPs and 79 graph coloring problems from the DIMACS challenge to test the performance of MAEA-CSPs for nonpermutation CSPs. MAEA-CSPs is compared with six well-defined algorithms and the effect of the parameters is analyzed systematically. The second part of the experiments uses a classical CSP, n-queen problems, and a more practical case, job-shop scheduling problems (JSPs), to test the performance of MAEA-CSPs for permutation CSPs. The scalability of MAEA-CSPs along n for n-queen problems is studied with great care. The results show that MAEA-CSPs achieves good performance when n increases from 10(4) to 10(7), and has a linear time complexity. Even for 10(7)-queen problems, MAEA-CSPs finds the solutions by only 150 seconds. For JSPs, 59 benchmark problems are used, and good performance is also obtained.
    IEEE TRANSACTIONS ON CYBERNETICS 03/2006; 36(1):54-73. DOI:10.1109/TSMCB.2005.852980 · 6.22 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: There are several evolutionary approaches for solving ran-dom binary Constraint Satisfaction Problems (CSPs). In most of these strategies we find a complex use of informa-tion regarding the problem at hand. Here we present a hy-brid Evolutionary Algorithm that outperforms previous ap-proaches in terms of effectiveness and compares well in terms of efficiency. Our algorithm is conceptual and simple, fea-turing a GRASP-like (GRASP stands for Greedy Random-ized Adaptive Search Procedure) mechanism for genotype-to-phenotype mapping, and without considering any specific knowledge of the problem. Therefore, we provide a simple algorithm that harnesses generality while boosting perfor-mance.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: La conception d'avions au stade avant-projet consiste à déterminer les principales caractéristiques d'un avion répondant à un cahier des charges donné. Ces études peuvent être résumées par des problèmes d'optimisation globale sous contraintes avec typiquement un millier de paramètres et presque autant de contraintes. Les contraintes expriment la faisabilité physique ainsi que le cahier des charges à respecter, et les objectifs sont des performances de l'avion guidées par des études de marché. De plus, le conception d'avions est un problème d'optimisation multicritère du fait de la présence de fonctions objectifs antagonistes. L'objectif de cette thèse est d'introduire de nouvelles méthodes mathématiques qui peuvent être utiles dans un outil de dimensionnement avant-projet pour résoudre le problème d'optimisation d'une configuration d'avion. Nous avons contribué à l'amélioration des méthodes d'optimisation qui sont couramment utilisées au département des Avant-Projets d'Airbus. En utilisant les algorithmes génétiques, nous avons rendu le processus d'optimisation monocritère plus robuste. Ensuite, nous avons introduit des méthodes d'optimisation multicritère car nous avions plusieurs critères conflictuels à considérer. Comme les temps de calcul sont devenus importants, nous avons décidé de substituer au modèle d'avion un modèle approché. Nous avons implémenté les fonctions à base radiale pour approcher les contraintes et les fonctions objectifs. Enfin, nous avons propagé les incertitudes du modèle pour estimer la robustesse des résultats de l'optimisation et nous avons proposé un aboutissement possible de l'intégration de ces techniques : donner aux ingénieurs une perception opérationnelle de l'espace de définition.
Show more