Mahfoud, S. W.: Crowding and Preselection Revisited. In: Manner, R. and Manderick, B (eds.): Parallel Problem Solving from Nature, 2, Amsterdam, Elsevier Science (1992) 27-36  Home/Search   Document Details and Download   Summary   Related Articles   Check

....as well as local optima in a multimodal domain. As a result, several population diversity mechanisms have been proposed to counteract the convergence of the population to a single solution by maintaining a diverse population of members throughout its search. These methods, known as niching methods [12, 16, 17, 18], were designed to identify multiple optima within multimodal domains. Each peak in a mutlimodal domain can be thought of as a niche. In nature, niches correspond to different subspaces of the environment that can support different types of life such as species or organisms. The fertility of the ....

....is what determines the number of organisms that can be contained in a niche. This principle is at the base of how GAs should maintain the population diversity of its members in a multimodal domain. Thus, the niches should be populated in proportion to their fitness relative to other peaks. Mahfoud [18] proposed an improved crowding mechanism, called deterministic crowding (DC) After the mating of 2 parents, DC replaces each parent by the most similar child only if the latter has higher fitness. 1.4 Gene Expression in Nature Dramatic tansformations in nature occur when plants and animals ....

S. W. Mahfoud, "Crowding and preselection revisited," in Parallel problem Solving from Nature, PPSN '92, Brussels, 1992.

....genetic differentiation plus the selection force from the environment. The idea of speciation is borrowed byGAcommunity to cater for the problems with landscapes that are multi modal or deceptiveinnature. Anumber of speciation algorithms were proposed, namely, crowding [11] deterministic crowding [42], sharing [28,43] and dynamic niching [44] 2.3 Crowding methods Crowding Crowding [11] is a competition model to maintain population diversity to prevent premature convergence rather than a species formation method. It is an extension to the preselection method of Cavicchio [9] ....

....Although less fit (and similar) individuals are kept replacing, only mild speciation effect is produced and it is empirically shown to be not effective in solving multimodal problems [12] The computational complexity of this method is O(nm) 2. 3 Deterministiccrowding Developed by Mahfond [42]toimprove the high computational requirementofDeJong s crowding method. One of the weaknesses of DeJong s crowding method is the expensive computation used in the replacement of parents by offsprings. The replacement method of deterministic crowding is that only parents and their direct ....

[Article contains additional citation context not shown here]

S.W. Mahfoud. Crowding and preselection revisited. In R. Manner and B Manderick, editors, Parallel problem solving from nature, 2: Proceedings of the 2nd ConferenceonParallel Problem Solving from Nature, Brussels, Belgium, 28-30 Sept

....do not restrict replacement as strongly as the island and the spatial models do; they involve a high degree of randomization in the choice of which population member a new genome should replace. Since this decision is randomized, sampling error can a ect replacement, causing genetic drift. Mahfoud (1992) cites this stochastic error as a diculty with crowding and preselection. As expected, those speciation methods that maintained the highest diversity also provided the best advantage for con dence evolution. Populations evolved using the island model and the spatial model were diverse enough so ....

Mahfoud, S. (1992). Crowding and preselection revisited. Parallel Problem Solving from Nature, 2:27-36.

....niche may have a niche radius independent of other niches in the problem space. There are other techniques that do not employ a radius as such, and therefore are, perhaps, not nichers in the true sense of the word, but are worthy of note. In particular are tagging [13] and deterministic crowding [11]. For the sake of brevity the specific minutiae of all these mechanisms have been omitted and the reader is referred to the citations for further details. There are many more methodologies and techniques designed to preserve diversity and encourage speciation in publication that are too numerous ....

Mahfoud, S.W.: Crowding and Preselection Revisited, In R.Manner & B.Manderick (eds), Parallel Problem Solving From Nature, 2, pp27-36, Elsevier Science Publishers B.V., 1992.

....solutions, such as in an LCS, where the GA searches for a set of rules which together implement a successful classification strategy. A number of niching mechanisms have been proposed and used over the last couple of decades. One of the earliest was Cavicchio s preselection (Cavicchio, 1970; Mahfoud, 1992), in which offspring could only replace one of their parents. DeJong s crowding (DeJong, 1975; Mahfoud, 1992) had the same flavor, in that new individuals replaced less fit, but similar, solutions in the old population. Boltzmann tournament selection has also been shown to have niching effects ....

....classification strategy. A number of niching mechanisms have been proposed and used over the last couple of decades. One of the earliest was Cavicchio s preselection (Cavicchio, 1970; Mahfoud, 1992) in which offspring could only replace one of their parents. DeJong s crowding (DeJong, 1975; Mahfoud, 1992) had the same flavor, in that new individuals replaced less fit, but similar, solutions in the old population. Boltzmann tournament selection has also been shown to have niching effects (Mahfoud, 1991) and recently immune system models (Smith, Forrest, Perelson, 1992 ) have been gaining ....

Mahfoud, S. W. (1992). Crowding and preselection revisited. Parallel Problem Solving From Nature, 2. North-Holland, 1992, pp. 27-36.

....algorithm into a crowding algorithm called the elitist crowding algorithm . The standard crowding algorithm randomly selects a subset of the population and replaces the string that matches the child most closely, whether or not the child s fitness value is better. Deterministic crowding (Mahfoud, 1992) and elitist recombination let the children compete only with their parents, but they need to have a better fitness value to enter the population. Generalizing the subset selection mechanism and the local elitism principle, we obtain an elitist crowding algorithm that captures the characteristics ....

Mahfoud, S. W. (1992). Crowding and preselection revisited. In M anner, R. and Manderick, B., editors, Proceedings of Parallel Problem Solving from Nature PPSN-II, pages 27--36, North-Holland, Amsterdam, The Netherlands.

....John, 444] Lovell, Byrne, 445] Lucasius, Carlos B. 446, 447, 448, 449, 450] Lybanon, M. 491] Machado, Ricardo Jose, 715] Maclay, David, 457] MacLennan, Bruce J. 456] Maeda, Y. 370] Magdalena, Luis, 675] Magele, C. A. 268] 16 Genetic algorithms of 1992 Mahfoud, Samir W. [250, 261, 458, 459, 460] Maimon, O. 532] Manderick, Bernard, 461] Mandischer, Martin, 462] Manela, Mauro, 342] Maniezzo, Vittorio, 166, 171, 463, 464] Manner, R. 315] Margarita, Sergio, 465] Marici c, Borut, 466] Marin, Francisco Javier, 615] Marko, K. A. 288] Markowicz, Bernard P. 135] Marks, ....

.... 300, 303] SAT, 20, 27] scheduling, 26, 38, 62, 97, 111, 126, 132, 187, 188, 190, 338, 372, 397, 433, 543, 547, 581, 604, 657, 663, 674, 18] scheduling manufacturing, 524] schema, 138] schema variance, 540] search, 41, 120, 279, 280] local, 42] seismology, 453, 612] selection, [434, 446, 458, 459, 473] interactive, 565] self organization, 54] SGA, 356] shape genes, 719] signal processing, 133, 134, 244, 316, 351, 429, 491, 503, 664] simulated annealing, 98, 261, 319, 320, 339, 367, 467] simulated evolution, 202] simulation, 407, 408, 648] communication, 236] SISAL, 122] ....

[Article contains additional citation context not shown here]

Samir W. Mahfoud. Crowding and preselection revisited. In Manner and Manderick [539], pages 27--36. y ga:Mahfoud92b.

.... tune the selective pressure 2 Evolutionary Computation Volume 8, Number EVOLUTIONARY LOCAL SELECTION ALGORITHMS adaptively, by a nonlinear scaling of the fitness function (Menczer and Parisi, 1992) The most notable selection variations explicitly aimed at niching are crowding (De Jong, 1975; Mahfoud, 1992) and fitness sharing (Goldberg and Richardson, 1987; Horn, 1993; Mahfoud, 1993) In both of these methods selection is somehow altered to take into account some measure of similarity among individuals. Shortcomings of both methods, as well as of other multi objective EAs, are inefficiency and ....

Mahfoud, S. (1992). Crowding and preselection revisited. In Proc. Parallel Problem Solving from Nature 2.

....scaling of the fitness function [47] Different selection methods of course impose varying degrees of selection pressure. For example, tournament selection is known to converge slowly and to have niching effects [15] The most notable selection variations explicitly aimed at niching are crowding [9, 31] and fitness sharing [17, 23, 32] In both of these methods, selection is somehow altered to take into account some measure of similarity among individuals. Shortcomings of both methods are problem dependency and inefficiency; if # is the population size, selection with sharing or crowding has ....

SW Mahfoud. Crowding and preselection revisited. In Parallel Problem Solving from Nature 2, 1992.

....Pareto optimality ranking method outperformed the VEGA method. The Pareto method was found to be superior to a VEGA by Liepins et al. 1990] when applied to a variety of set covering problems. Ritzel et al. 1994] also used non dominated ranking and selection combined with deterministic crowding [Mahfoud 1992] as the niching mechanism. They applied the GA to a groundwater pollution containment problem in which cost and reliability were the objectives. Though the actual Pareto front was unknown, Ritzel et al. used the best trade off surface found by a domain specific algorithm, called MICCP (Mixed ....

Mahfoud, S. M. 1992. Crowding and preselection revisited. In R. M anner and B. Manderick Eds., Parallel problem Solving from Nature, 2nd Workshop (Amsterdam, 1992). North-Holland Publishing Company.

....the probability of premature convergence. Within our implementation each offspring competes with only one of its parents. This scheme has a lower selective pressure than the standard elitist recombination scheme. This competition with one parent is also used in the deterministic crowding scheme [4]. But deterministic crowding lets offspring compete with the most similar parent. The Elitist recombination algorithm is chosen as it does not deteriorate the average fitness when inferior offspring is produced due to the (population) elitism, and it prevents too rapid convergence of the ....

S.W. Mahfoud. Crowding and preselection revisited. In Parallel Problem Solving from Nature II, pages 27--36. Springer, 1992.

....Both are meant to preserve the diversity in the population. It has been shown however that in practice stochastic errors in the replacement of population members work to create a significant amount of genetic drift, causing unsuccesfulness at preserving population diversity as noted by Mahfoud [27]. At this point we are not going to take a closer look at reasons or experimental explanations to confirm these results in a scientifically justified way. What we will do however is briefly go over the crowding and the preselection scheme as well as the scheme that was introduced by Mahfoud [27] ....

....[27] At this point we are not going to take a closer look at reasons or experimental explanations to confirm these results in a scientifically justified way. What we will do however is briefly go over the crowding and the preselection scheme as well as the scheme that was introduced by Mahfoud [27] to overcome problems introduced by crowding and preselection. Subsequently we shall look at the most popular way in GA history to achieve multimodal function optimization, which is sharing. Crowding The crowding strategy makes use of a crowding factor CF which is the number of genomes from the ....

[Article contains additional citation context not shown here]

S. Mahfoud. Crowding And Preselection Revisited. In: Parallel Problem Solving From Nature II -- PPSN II (R. Manner, B. Manderick, eds.), Springer--Verlag, Berlin, 1992, pp. 27--36.

.... Segrest, 1987) even if a problem has multiple solutions of equivalent fitness. Unfortunately, a single rule is usually insufficient to represent the desired concepts in either the financial domain or other complex domains. A set of interacting rules is required. GAs that employ niching methods (Mahfoud, 1992, 1995a, 1995b) are capable of finding and maintaining multiple rules using a single population. The basic idea behind niching is to simultaneously optimize multiple regions of the search space by reducing competition among sufficiently dissimilar individuals. In financial prediction, different ....

....1987) is one type of niching method that has been applied to classification (Booker, 1982; Horn et al. 1994; Kargupta Smith, 1991; Packard, 1990; Smith Valenzuela Rendn, 1989) Fitness sharing works by reducing the fitnesses of similar population elements. Crowding (De Jong, 1975; Mahfoud, 1992, 1994, 1995a) is another type of niching method that has also been applied to classification (Booker, 1982; Goldberg, 1983; Holland Reitman, 1978; Sedbrook et al. 1991; Stadnyk, 1987) Crowding forces newly generated population elements to replace older elements that are similar. Sequential ....

Mahfoud, S. W. 1992. Crowding and preselection revisited. In Parallel problem solving from nature, 2, eds. R. Mnner and B. Manderick, 2736. Amsterdam: Elsevier.

....the strings with all 1 s will soon take over, as all end bias is eliminated. Double acceptance rejection has a similar effect. Another possibility exists, to always match each parent with the closest child. This will encourage the maintenance of multiple optima, even at minimal temperatures [32]. The issues of diversity and convergence under these alternative competitions deserve further examination. In summary, the end biases can be ignored since PRSA maintains diversity via higher temperatures and mutation, they can be eliminated through random pairing or double acceptance rejection, ....

S. W. Mahfoud, Crowding and preselection revisited, in: R. Manner and B. Manderick, eds., Parallel Problem Solving from Nature, 2 (Elsevier, Amsterdam, 1992) 27--36.

....optimization problems, including a multimodal deceptive problem. I. Introduction Niching methods in genetic algorithms (GAs) strive to locate and maintain stable subpopulations or niches, within a single population. While sharing techniques [1, 2] have proven effective, crowding methods [3, 4] are also promising. Previously, little was known about the expected behavior of crowding methods. In this study, crowding turns out to behave far differently than sharing. Intrinsic properties of crowding are uncovered via controlled experimentation and intuitive modelling. Specifically, we ....

....one. To approximate this, a sample is taken from the existing population, and the sample element closest in Hamming distance to the new one is replaced. Unfortunately, stochastic replacement errors prevent De Jong s crowding from maintaining more than two peaks of a multimodal fitness landscape [3]. An improved crowding algorithm, deterministic crowding (DC) is developed in [3] DC nearly eliminates replacement errors, and is effective at locating and maintaining multiple niches. Figure 1 compares one run of De Jong s crowding with one run of deterministic crowding on F1, a five peak sine ....

[Article contains additional citation context not shown here]

S. W. Mahfoud, "Crowding and preselection revisited", Proc. 2nd Conf. Parallel Problem Solving from Nature, PPSN '92, Brussels, Belgium, Sep. 28--30, 1992 (Elsevier).

No context found.

Mahfoud, S. W.: Crowding and Preselection Revisited. In: Manner, R. and Manderick, B (eds.): Parallel Problem Solving from Nature, 2, Amsterdam, Elsevier Science (1992) 27-36

No context found.

S. W. Mahfoud, "Crowding and preselection revisited," in Proceedings Second Conference Parallel Problem Solving from Nature, 1992.

No context found.

S.W . Mahfoud, "Crowding and preselection re visited,"Parallel Problem Solving from Nature 2, Brussels Belgium pp. 27-36 1992.

No context found.

S. W. Mahfoud, "Crowding and preselection revisited," in Proceedings Second Conference Parallel Problem Solving from Nature, 1992.

No context found.

S. W. Mahfoud. Crowding and preselection revisited. In R. Manner and B. Manderick, editors, Parallel Problem Solving from Nature II, pages 27--36, Amsterdam, 1992. North-Holland.

No context found.

S. W. Mahfoud. Crowding and preselection revisited. In Reinhard Manner and Bernard Manderick, editors, Parallel problem solving from nature 2, pages 27--36, Amsterdam, 1992. North-Holland.

No context found.

S. W. Mahfoud. Crowding and preselection revisited. In R. Manner and B. Manderick, editors, Parallel Problem Solving from Nature II, pages 27--36, Amsterdam, 1992. North-Holland.

No context found.

Mahfoud, S., "Crowding and Preselection Revisited," Parallel Problem Solving from NaturePPSN2, edited by R. Manner and B. Manderick, Elsevier Science Publishers B. V., 1992, pp. 27-36.

No context found.

S. W. Mahfoud. Crowding and preselection revisited. In R. Manner and B. Manderick, editors, Parallel Problem Solving from Nature, 2, pages 27--36. North-Holland, 1992.

No context found.

S. W. Mahfoud. Crowding and preselection revisited. In R. Manner and B. Manderick, editors, Parallel Problem Solving from Nature, 2, pages 27--36. 1992.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC