Showing papers presented at "Congress on Evolutionary Computation in 2000"

Comparing inertia weights and constriction factors in particle swarm optimization

Abstract: The performance of particle swarm optimization using an inertia weight is compared with performance using a constriction factor. Five benchmark functions are used for the comparison. It is concluded that the best approach is to use the constriction factor while limiting the maximum velocity Vmax to the dynamic range of the variable Xmax on each dimension. This approach provides performance on the benchmark functions superior to any other published results known by the authors.

Genetic quantum algorithm and its application to combinatorial optimization problem

Abstract: This paper proposes a novel evolutionary computing method called a genetic quantum algorithm (GQA). GQA is based on the concept and principles of quantum computing such as qubits and superposition of states. Instead of binary, numeric, or symbolic representation, by adopting qubit chromosome as a representation GQA can represent a linear superposition of solutions due to its probabilistic representation. As genetic operators, quantum gates are employed for the search of the best solution. Rapid convergence and good global search capability characterize the performance of GQA. The effectiveness and the applicability of GQA are demonstrated by experimental results on the knapsack problem, which is a well-known combinatorial optimization problem. The results show that GQA is superior to other genetic algorithms using penalty functions, repair methods and decoders.

On measuring multiobjective evolutionary algorithm performance

Abstract: Solving optimization problems with multiple (often conflicting) objectives is generally a quite difficult goal. Evolutionary algorithms (EAs) were initially extended and applied during the mid-eighties in an attempt to stochastically solve problems of this generic class. During the past decade a multiplicity of multiobjective EA (MOEA) techniques have been proposed and applied to many scientific and engineering applications. Our discussion's intent is to rigorously define and execute a quantitative MOEA performance comparison methodology. Almost all comparisons cited in the current literature visually compare algorithmic results, resulting in only relative conclusions. Our methodology gives a basis for absolute conclusions regarding MOEA performance. Selected results from its execution with four MOEAs are presented and described.

M-PAES: a memetic algorithm for multiobjective optimization

Abstract: A memetic algorithm for tackling multiobjective optimization problems is presented. The algorithm employs the proven local search strategy used in the Pareto archived evolution strategy (PAES) and combines it with the use of a population and recombination. Verification of the new M-PAES (memetic PAES) algorithm is carried out by testing it on a set of multiobjective 0/1 knapsack problems. On each problem instance, a comparison is made between the new memetic algorithm, the (1+1)-PAES local searcher, and the strength Pareto evolutionary algorithm (SPEA) of E. Zitzler and L. Thiele (1998, 1999).

Stereotyping: improving particle swarm performance with cluster analysis

Abstract: Individuals in the particle swarm population were "stereotyped" by cluster analysis of their previous best positions. The cluster centers then were substituted for the individuals' and neighbors' best previous positions in the algorithm. The experiments, which were inspired by the social-psychological metaphor of social stereotyping, found that performance could be generally improved by substituting individuals', but not neighbors', cluster centers for their previous bests.

Evolutionary artificial potential fields and their application in real time robot path planning

Abstract: A new methodology named Evolutionary Artificial Potential Field (EAPF) is proposed for real-time robot path planning. The artificial potential field method is combined with genetic algorithms, to derive optimal potential field functions. The proposed EAPF approach is capable of navigating robot(s) situated among moving obstacles. Potential field functions for obstacles and goal points are also defined. The potential field functions for obstacles contain tunable parameters. The multi-objective evolutionary algorithm (MOEA) is utilized to identify the optimal potential field functions. Fitness functions such as goal-factor, obstacle-factor, smoothness-factor and minimum-pathlength-factor are developed for the MOEA selection criteria. An algorithm named escape-force is introduced to avoid the local minima associated with EAPF. Moving obstacles and moving goal positions were considered to test the robust performance of the proposed methodology. Simulation results show that the proposed methodology is efficient and robust for robot path planning with non-stationary goals and obstacles.

Handling preferences in evolutionary multiobjective optimization: a survey

Abstract: Despite the relatively high volume of research conducted on evolutionary multiobjective optimization in the last few years. Little attention has been paid to the decision making process that is required to select a final solution to the multiobjective optimization problem at hand. This paper reviews the most important preference handling approaches used with evolutionary algorithms, analyzing their advantages and disadvantages, and then, it proposes some of the potential areas of future research in this discipline.

Convergence properties of some multi-objective evolutionary algorithms

Abstract: We present four abstract evolutionary algorithms for multi-objective optimization and theoretical results that characterize their convergence behavior. Thanks to these results it is easy to verify whether or not a particular instantiation of these abstract evolutionary algorithms offers the desired limit behavior. Several examples are given.

Discovering comprehensible classification rules with a genetic algorithm

Abstract: Presents a classification algorithm based on genetic algorithms (GAs) that discovers comprehensible IF-THEN rules, in the spirit of data mining. The proposed GA has a flexible chromosome encoding, where each chromosome corresponds to a classification rule. Although the number of genes (the genotype) is fixed, the number of rule conditions (the phenotype) is variable. The GA also has specific mutation operators for this chromosome encoding. The algorithm was evaluated on two public-domain real-world data sets (in the medical domains of dermatology and breast cancer).

A unified model for multi-objective evolutionary algorithms with elitism

Abstract: Though it has been claimed that elitism could improve evolutionary multi-objective search significantly, a thorough and extensive evaluation of its effects is still missing. Guidelines on how elitism could successfully be incorporated have not yet been developed. This paper presents a unified model of multi-objective evolutionary algorithms, in which arbitrary variation and selection operators can be combined as building blocks, including archiving and re-insertion strategies. The presented model enables most specific multi-objective (evolutionary) algorithm to be formulated as an instance of it, which will be demonstrated by simple examples. We further show how elitism can be quantified by the model's parameters and how this allows an easy evaluation of the effect of elitism on different algorithms.

A multiobjective evolutionary setting for feature selection and a commonality-based crossover operator

Abstract: Feature selection is a common and key problem in many classification and regression tasks. It can be viewed as a multiobjective optimisation problem, since, in the simplest case, it involves feature subset size minimisation and performance maximisation. This paper presents a multiobjective evolutionary approach for feature selection. A novel commonality-based crossover operator is introduced and placed in the multiobjective evolutionary setting. This specialised operator helps to preserve building blocks with promising performance. Selection bias reduction is achieved by resampling. We argue that this is a generic approach, which can be used in many modelling problems. It is applied to feature selection on different neural network architectures. Results from experiments with benchmarking data sets are given.

A multiobjective genetic algorithm for radio network optimization

Abstract: Engineering of mobile telecommunication networks endures two major problems: the design of the network and the frequency assignment. We address the first problem in this paper, which has been formulated as a multiobjective constrained combinatorial optimisation problem. We propose a genetic algorithm (GA) that aims to approximate the Pareto frontier of the problem. Advanced techniques have been used, such as Pareto ranking, sharing and elitism. The GA has been implemented in parallel on a network of workstations to speed up the search. To evaluate the performance of the GA, we have introduced two new quantitative indicators: the entropy and the contribution. Encouraging results are obtained on real-life problems.

Nash genetic algorithms: examples and applications

Abstract: This article presents both theoretical aspects and experimental results for Nash genetic algorithms. Nash GAs are an alternative for multiple objective optimization as they are an optimization tool based on noncooperative game theory. They are explained in detail, along with the advantages conferred by their equilibrium state. This approach is tested on a few benchmark problems, and some comparisons are made with Pareto GAs, particularly in terms of speed and robustness. The different concepts presented in this paper are then illustrated via experiments on a computational fluid dynamics problem, namely nozzle reconstruction with multiple criteria (subsonic and transonic shocked flows). The overall results are that Nash genetic algorithms offer a fast and robust alternative for multiple objective optimization.

An efficient evolutionary algorithm for the degree-constrained minimum spanning tree problem

Abstract: The representation of candidate solutions and the variation operators are fundamental design choices in an evolutionary algorithm (EA). This paper proposes a novel representation technique and suitable variation operators for the degree-constrained minimum spanning tree problem. For a weighted, undirected graph G(V, E), this problem seeks to identify the shortest spanning tree whose node degrees do not exceed an upper bound d/spl ges/2. Within the EA, a candidate spanning tree is simply represented by its set of edges. Special initialization, crossover, and mutation operators are used to generate new, always feasible candidate solutions. In contrast to previous spanning tree representations, the proposed approach provides substantially higher locality and is nevertheless computationally efficient; an offspring is always created in O(|V|) time. In addition, it is shown how problem-dependent heuristics can be effectively incorporated into the initialization, crossover, and mutation operators without increasing the time-complexity. Empirical results are presented for hard problem instances with up to 500 vertices. Usually, the new approach identifies solutions superior to those of several other optimization methods within few seconds. The basic ideas of this EA are also applicable to other network optimization tasks.

Triggered hypermutation revisited

Abstract: With the emergence of standardized problem generators for dynamic problem environments, we are just starting to systematically measure the performance of different evolutionary-algorithm (EA) extensions against standard classes of problems. We revisit triggered hypermutation, one of the early and most successful implementations of EA's for dynamic environments. Using an implementation of this algorithm, we systematically evaluate the performance of triggered hypermutation on specific test problems across a range of values for the environmental change rate relative to the EA "time" measured in generations. We examine the results, identify a probable cause for the algorithm's behavior, and suggest some improvements to the algorithm.

The new model of parallel genetic algorithm in multi-objective optimization problems - divided range multi-objective genetic algorithm

Abstract: Proposes a divided-range multi-objective genetic algorithm (DRMOGA), which is a model for the parallel processing of genetic algorithms (GAs) for multi-objective problems. In the DRMOGA, the population of GAs is sorted with respect to the values of the objective function and divided into sub-populations. In each sub-population, a simple GA for multi-objective problems is performed. After some generations, all the individuals are gathered and they are sorted again. In this model, the Pareto-optimal solutions which are close to each other are collected into one sub-population. Therefore, this algorithm increases the calculation efficiency and a neighborhood search can be performed. Through numerical examples, the following facts become clear: (i) the DRMOGA is a very suitable GA model for parallel processing, and (ii) in some cases it can derive better solutions compared to both the single-population model and the distributed model.

An investigation of redundant genotype-phenotype mappings and their role in evolutionary search

Abstract: The neutral theory of evolution suggests that most mutations do not cause a phenotypic change. In this case the mapping from genotype to phenotype contains redundancy such that many mutations do not have an appreciable effect on the phenotype. This can result in neutral networks; sets of genotypes connected by single point mutations that map to the same phenotype. A population is able to drift along these networks, eventually encountering phenotypes of higher fitness, thus reducing the chance of becoming trapped in sub-optimal regions of genotype space. In this paper we explore the use and benefit of redundant mappings for evolutionary search. We investigate the properties of several genotype-phenotype mappings by performing random walks along the neutral networks in their genotype spaces. The properties are explored further by performing adaptive walks in which a concept of fitness is introduced. A mapping based on a random Boolean network was found to have particularly interesting properties in both cases.

Towards automatic image enhancement using genetic algorithms

Abstract: This paper introduces a new automatic image enhancement technique based on real-coded genetic algorithms (GAs). The task of the GA is to adapt the parameters of a novel extension to a local enhancement technique similar to statistical scaling, as to enhance the contrast and detail in the image according to an objective fitness criterion. We compared our method with other automatic enhancement techniques, like contrast stretching and histogram equalization methods. Results obtained, both in terms of subjective and objective evaluation, show the superiority of our method.

An extended genetic rule induction algorithm

Abstract: Describes an extension of a genetic algorithm (GA) based separate-and-conquer propositional rule induction algorithm called SIA (Supervised Inductive Algorithm). While the original algorithm is computationally attractive and is also able to handle both nominal and continuous attributes efficiently, our algorithm further improves it by taking into account recent advances in the rule induction and evolutionary computation communities. The refined system has been compared to other GA-based and non-GA-based rule learning algorithms on a number of benchmark data sets from the UCI (University of California, Irvine) machine learning repository. The results show that the proposed system can achieve higher performance while still producing a smaller number of rules.

Evolutionary algorithms for nurse scheduling problem

Abstract: The nurse scheduling problem (NSPs) represents a difficult class of multi-objective optimisation problems consisting of a number of interfering objectives between the hospitals and individual nurses. The objective of this research is to investigate difficulties that occur during the solution of NSP using evolutionary algorithms, in particular genetic algorithms (GA). As the solution method a population-less cooperative genetic algorithm (CGA) is taken into consideration. Because contrary to competitive GAs, we have to simultaneously deal with the optimization of the fitness of the individual nurses and also optimization of the entire schedule as the final solution to the problem in hand. To confirm the search ability of CGA, first a simplified version of NSP is examined. Later we report a more complex and useful version of the problem. We also compare CGA with another multi-agent evolutionary algorithm using pheromone style communication of real ants. Finally, we report the results of computer simulations acquired throughout the experiments.

An evolvable hardware FPGA for adaptive hardware

Abstract: Can we realise the opportunities that lie in design by evolution by using traditional technologies or are there better technologies which will allow us to fully realise the potential inherent in evolvable hardware? The authors consider the characteristics of evolvable hardware, especially for adaptive design, and discuss the demands that these characteristics place on the underlying technology. They suggest a potential alternative to today's FPGA technology. The proposed architecture is particularly focused at reducing the genotype required for a given design by reducing the configuration data required for unused routing resources and allowing partial configuration down to a single CLB. In addition, to support adaptive hardware, self-reconfiguration is enabled.

A genetic algorithm for the orienteering problem

Abstract: This paper presents a genetic algorithm to solve the orienteering problem, which is concerned with finding a path between a given set of control points, among which a start and an end point are specified, so as to maximize the total score collected subject to a prescribed time constraint. Employing three sets of test problems from the literature, the performance of the genetic algorithm is evaluated against problem specific heuristics and an artificial neural network.

Genetic programming polynomial models of financial data series

Abstract: The problem of identifying the trend in financial data series in order to forecast them for profit increase is addressed using genetic programming (GP). We enhance a GP system that searches for polynomial models of financial data series and relate it to a traditional GP manipulating functional models. Two of the key issues in the development are: 1) preprocessing of the series which includes data transformations and embedding; and 2) design of a proper fitness function that navigates the search by favouring parsimonious and predictive models. The two GP systems are applied for stock market analysis, and examined with real Tokyo Stock Exchange data. Using statistical and economical measures to estimate the results, we show that the GP could evolve profitable polynomials.

Anaconda defeats Hoyle 6-0: a case study competing an evolved checkers program against commercially available software

Abstract: We have been exploring the potential for a coevolutionary process to learn how to play checkers without relying on the usual inclusion of human expertise in the form of features that are believed to be important to playing well. In particular, we have focused on the use of a population of neural networks, where each network serves as an evaluation function to describe the quality of the current board position. After only a little more than 800 generations, the evolutionary process has generated a neural network that can play checkers at the expert level as designated by the US Chess Federation rating system. The current effort reports on a competition between the best-evolved neural network, named "Anaconda," and commercially available software. In a series of six games, Anaconda scored a perfect six wins.

Evolutionary computation techniques for multiple sequence alignment

Abstract: Given a collection of biologically related protein or DNA sequences, the basic multiple sequence alignment problem is to determine the most biologically plausible alignment of these sequences. Under the assumption that the collection of sequences arose from some common ancestor, an alignment can be used to infer the evolutionary history among the sequences, i.e., the most likely pattern of insertions, deletions and mutations that transformed one sequence into another. The general multiple sequence alignment problem is known to be NP-hard, and hence the problem of finding the best possible multiple sequence alignment is intractable. However, this does not preclude the possibility of developing algorithms that produce near optimal multiple sequence alignments in polynomial time. We examine techniques to combine efficient algorithms for near optimal global and local multiple sequence alignment with evolutionary computation techniques to search for better near optimal sequence alignments. We describe our evolutionary computation approach to multiple sequence alignment and present preliminary simulation results on a set of 17 clusters of orthologous groups of proteins (COGs). We compare the fitness of the alignments given by the proposed techniques with the fitness of CLUSTAL W alignments given in the COG database.

Parameter estimation of a bioreaction model by hybrid differential evolution

Abstract: Hybrid differential evolution is applied to estimate the kinetic model parameters of batch fermentation for ethanol and glycerol production using Saccharomyces diastaticus LORRE 316. In this study, we consider the worst observed error for all experiments as an objective function so that the parameter estimation problem becomes a min-max estimation problem. Several methods have been employed to solve the min-max estimation problem for comparison. The proposed method can use a small population size to obtain a more satisfactory solution as compared from these computations. In order to validate the kinetic model, we have carried out the fedbatch experiments with an optimal feed rate. The experimental data can fit the computed results satisfactorily.

Evolving pet robot with emotional model

Abstract: Deals with a pet robot with an emotional model. The robot requires several capabilities, such as perceiving, acting, communicating and surviving. Furthermore, it should learn various behaviors through interaction with its owner. This paper focuses on teaching a pet robot tricks or to dance. Basically, the owner can teach these tricks by simple communication based on trial and error. The robot performs the tricks by using a fuzzy controller, and further acquires tricks by a delta rule for online learning and a genetic algorithm for off-line learning. We use "Rag Warrior" as our pet robot. Experimental results show that this robot performs tricks through interaction with its owner.

Ant colonies are good at solving constraint satisfaction problems

Abstract: We define an ant algorithm for solving random binary constraint satisfaction problems (CSPs). We empirically investigate the behavior of the algorithm on this type of problems and establish the parameter settings under which the ant algorithm performs best for a specific class of CSPs. The ant algorithm is compared to six other state-of-the-art stochastic algorithms from the field of evolutionary computing. It turns out that the ant algorithm outperforms all other algorithms and that bivariate distribution algorithms perform worse than the univariate ones, the latter largely due to the fact that they cannot model the randomly generated instances.

A game-theoretic investigation of selection methods used in evolutionary algorithms

Abstract: The replicator equation used in evolutionary game theory (EGT) assumes that strategies reproduce in direct proportion to their payoffs; this is akin to the use of fitness-proportionate selection in an evolutionary algorithm (EA). In this paper, we investigate how various other selection methods commonly used in EAs can affect the discrete-time dynamics of EGT. In particular, we show that the existence of evolutionary stable strategies (ESS) is sensitive to the selection method used. Rather than maintain the dynamics and equilibria of EGT, the selection methods we test either impose a fixed-point dynamic virtually unrelated to the payoffs of the game matrix, or they give limit cycles or induce chaos. These results are significant to the field of evolutionary computation because EGT can be understood as a coevolutionary algorithm operating under ideal conditions: an infinite population, noiseless payoffs and complete knowledge of the phenotype space. Thus, certain selection methods, which may operate effectively in simple evolution, are pathological in an ideal-world coevolutionary algorithm, and therefore dubious under real-world conditions.

A novel hybrid evolutionary programming method for function optimization

Abstract: The basic evolutionary programming (BEP) method utilizes individual parent fitness to generate offspring. This is objectionable in many optimization problems, where the fitness value grows rapidly with problem dimensions, and two optimization problems differ by simply a scale factor. This paper is concerned with the development of an evolutionary programming method, which is functionally, and structurally equivalent to BEP, but still can be used effectively to optimize functions having strong fitness dependency between parents and their offspring. In this paper, a fitness-blind mutation (FBM) algorithm has been proposed, and then this is used in conjunction with the BEP mutation operator. The FBM operation has been implemented by taking the standard deviation of the Gaussian variable to vary in proportion to the genotypic distance between the individual parent and the fittest individual, which is defined as a pseudo-global optimum individual in a population pool. Also, the directionality of the random variation has been exploited to improve the probability of getting better solutions. In addition to this, the importance of initial search width for generating the offspring has been established empirically. The effectiveness of the proposed algorithm has been verified on well-established test functions.