L. Booker. Recombination distribution for genetic algorithms. In Foundations of Genetic Algorithms 2, pages 29--44. Morgan Kaufmann, 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....gene is the meta one. This allows a complete description of the dynamical behavior of a one locus problem with two alleles in the basic search space This restriction may appear severe but much of the work in mathematical genetics assumes single locus models and large populations at equilibrium [3]. The distribution of genotypes, that is a population, is represented as usually by one point in the simplex f p 0 ; p 1 ;p 2 ;p 3 2 R 4 i p i = 1g, where p i represents the proportion of the individual i in the population. Duality imposes the equalities f 0 = f 3 and f 1 = f 2 (where f i ....

....feature of genetic algorithm; so the case of no selection is particularly important as a test of the power of genetic search. In this section, we use results from mathematical genetics: the theory of recombination that characterizes the e ects of recombination on multiple loci without selection [3, 13]. The probability that an individual has a 1 in a given locus remains constant under crossover is a fundamental property. That is, for any pair of parents, a crossover preserves the number of 1s at each position. So crossover does not change the distribution of alleles at any locus. S h u p0 ....

[Article contains additional citation context not shown here]

L. B. Booker. Recombination distributions for genetic algorithms. In L.D. Whitley, editor, Foundations of Genetic Algorithms 2, pages 29{ 44. Morgan Kaufmann, San Mateo, CA, 1993.

.... a Markov chain, developed by Nix and Vose [NV92] Davis independently developed a model based on Markov chains as well [Dav91b] Subsequently, Vose developed the model by Vose and Nix to further tie the genetic algorithm to the infinite population model [Vos91a] Some other work, notably Booker [Boo93] Whitley [Whi93] and Vose [Vos93] has also considered different models of genetic algorithms and aspects of genetic algorithms. These have centred on stochastic (or probabilistic) views of the genetic algorithm, and not emphasised the time aspect (apart from the role it naturally plays in e.g. ....

Lashon B. Booker. Recombination distributions for genetic algorithms. In Foundations of Genetic Algorithms 2, pages 29-- 44. Morgan Kaufmann, 1993.

....crossovers can simply be defined by selecting subsets of nodes in the common region differently. A good way to describe and model the class of homologous crossovers is to extend the notions of crossover masks and recombination distributions used in genetics [25] and in the GA literature [26, 27, 28]. In a GA operating on fixedlength strings a crossover mask is simply a binary string. When crossover is executed, the bits of the offspring corresponding to the 1 s in the mask will be taken from one parent, those corresponding to 0 s from the other parent. For example, if the parents are the ....

L. B. Booker, "Recombination distributions for genetic algorithms," in FOGA-92, Foundations of Genetic Algorithms, (Vail, Colorado), 24--29 July 1992. Email: booker@mitre.org.

....crossovers can simply be defined by selecting subsets of nodes in the common region differently. A good way to describe and model the class of homologous crossovers is to extend the notions of crossover masks and recombination distributions used in genetics [25] and in the GA literature [26, 27, 28]. In a GA operating on fixedlength strings a crossover mask is simply a binary string. When crossover is executed, the bits of the offspring corresponding to the 1 s in the mask will be taken from one parent, those corresponding to 0 s from the other parent. For example, if the parents are the ....

L. B. Booker, "Recombination distributions for genetic algorithms," in FOGA-92, Foundations of Genetic Algorithms, (Vail, Colorado), 24--29 July 1992. Email: booker@mitre.org.

.... and a xed point for the variance of the length distribution, see [Rowe and McPhee, 2001] 4 3 Geiringer s theorem In this section we brie y introduce Geiringer s theorem [Geiringer, 1944] an important result with implications both for natural population genetics and evolutionary algorithms [Booker, 1992, Booker et al. 2000, Spears, 2000] Geiringer s theorem indicates that, in a population of xed length chromosomes repeatedly undergoing crossover (in the absence of mutation and selective pressure) the probability of nding a generic string h 1 h 2 hN approaches a limit distribution ....

Booker, L. B. (1992). Recombination distributions for genetic algorithms. In FOGA-92, Foundations of Genetic Algorithms, Vail, Colorado. Email: booker@mitre.org.

....crossovers can simply be defined by selecting subsets of nodes in the common region differently. A good way to describe and model the class of homologous crossovers is to extend the notions of crossover mask and recombination distributions used in genetics (Geiringer 1944) and in the GA literature (Booker 1992, Altenberg 1995, Spears 2000) In a GA operating on fixed length strings a crossover mask is simply a binary string. When crossover is executed, the bits of the offspring corresponding to the 1 s in the mask will be taken from one parent, those corresponding to 0 s from the other parent. For ....

Booker, Lashon B. (1992). Recombination distributions for genetic algorithms. In: FOGA-92, Foundations of Genetic Algorithms. Vail, Colorado. Email: booker@mitre.org.

....is biased) An example of a biased crossover operator is the one point crossover. This operator is biased toward solutions involving schemata with a short defining length as these schemata are less likely to be disrupted by this operator. Biases in the crossover operators have been studied [Boo92, ECS89, SEO91]. We have developed transmission function models for a number of genetic algorithms [vK97a] Tracing these probabilistic models corresponds to running a genetic algorithm with an infinite population. The results obtained by tracing these models can be regarded as a bound on the reliability of the ....

L.B. Booker. Recombination distributions for genetic algorithms. In Foundations of Genetic Algorithms - 2, pages 29--44, 1992.

....up the next population. n Gamma1 = Gamma 1 n Gamma1 ; N n Gamma1 Delta Crossover Gamma Gamma Gamma Gamma Gamma Gamma Gamma Gamma Gamma Gamma Gamma Gamma n = Gamma 1 n ; N n Delta In the context of infinite populations, general results can be found in [46, 30] which prove that all complete 2 crossover schemes lead to the same limit distribution, given by Robbins proportions of individuals in the limiting population. x) Y i=1 p(x i ) 13) where p(x i ) is the constant marginal distribution of the allele at position i. Note that these ....

L. B. Booker. Recombination distributions for genetic algorithms. In L. D. Whitley, editor, Foundations of Genetic Algorithms 2, pages 29--44, Los Altos, CA, 1993. Morgan Kaufmann.

....Christian, 97] Bilbro, Griff L. 98] Biro, O. 268] Blanton, Jr. Joe L. 99] Blanton Jr. Joe L. 100] Blommers, Marcel J. J. 447] Blume, Christian, 266] Boers, Egber, 101] Bogardi, J. J. 102] Boggia, R. 434] Bohm, A. P. Wim, 693] Booker, A. 120] Booker, Lashon B. [103] Born, Joachim, 104, 105, 106, 107] Bornholdt, Stefan, 108] Borup, Liana, 544] Bowden, Royce O. 109] Bratko, Ivan, 191] Brill, Frank Z. 110] Brown, Donald E. 110, 135] Brown, J. 343] Bruns, Ralf, 111] Bukatova, Innesa L. 112, 113] Bullock, G. N. 545] Cai, H. 352] ....

.... Prolog, 440] proportional fitness, 360] protein folding, 59, 60, 360, 445, 447, 455, 507, 508, 620, 621, 15] psychology decision making models, 383] QAP, 213, 463, 528, 707] quality, 157] quality control, 308] random number generators, 32, 84] real coding, 245] recombination, [103, 299, 300, 572] discontinuous, 527] report of activities, 219] representations, 573, 574] response surface changing, 271] review, 19, 22, 70, 72, 74, 159, 178, 239, 1, 322, 323, 327, 335, 385, 395, 420, 461, 615, 695, 9] review artificial life, 431] Davidor, 25] GA and neural networks, 582, ....

[Article contains additional citation context not shown here]

Lashon B. Booker. Recombination distributions for genetic algorithms. In Whitley [11], pages 29--44. ga:Booker92a.

....only if it uses some knowledge about the problem. To tune a recombination operator to a problem s fitness landscape it is necessary to determine the features of that landscape. While there has been much research into recombination operators on genotypic encodings (see for example [11] 40] and [5]) visualizing the landscapes for genotypic EAs is very difficult. Thus we use a phenotypic representation as it is easier to identify the features of the landscape and tailor the recombination operator to them (both [38] and [27] also find the phenotypic space easier to work with) On a phenotypic ....

Lashon B. Booker. Recombination distributions for genetic algorithms. In L. D. Whitley, editor, Foundations of Genetic Algorithms 2, pages 29--44. Morgan Kaufmann, 1993.

....it considers only two forms of crossover, two point and uniform. Interestingly, although these two forms are very commonly used, they represent extremes. Two point crossover is the least disruptive of material, while uniform crossover is the most disruptive (De Jong and Spears 1992) Also, Booker (1992) has noted that, in terms of positional and distributional bias, both two point and uniform crossover are considerably different. Thus, it is natural to allow the GA to explore a relative mixture of these two operators, the motivation being that different mixtures will represent different ....

Booker, L. B. (1992). Recombination Distributions for Genetic Algorithms.

....on N . It is for N 2 n almost independent of n. In all simulations with n 4 the deviation from Robbins proportions was less than 1 after 7 generations. This strongly supports to analyze genetic algorithms by assuming Robbins proportions. The work of Geiringer was recently rediscovered by Booker (1993). He wrote: Geiringer s convergence results suggest that the most important difference among recombination operators is the rate at which they converge to equilibrium in the absence of selection. We will later show that it is difficult to extrapolate recombination results without selection to ....

Booker, L.B. (1993). Recombination distributions for genetic algorithms. In L.D.

....is affected by epistasis. These discussions frequently center on the usefulness, or lack thereof, of operators such as crossover and mutation in solving epistatic problems, with evidence (pro and con) in the form of theoretical arguments or empirical studies on a few carefully chosen examples (Booker 1992; Davidor 1990; De Jong and Spears 1992; Fogel 1995; Goldberg 1989; Holland 1975; Schaffer and Eshelman 1991; Spears 1992) The lack of emergence of a clear picture of these effects suggests to us that there is an opportunity here to improve our understanding by way of systematic studies using ....

Booker, L. B. (1992). Recombination distributions for genetic algorithms. In Proceedings of the Second Foundations of Genetic Algorithms Workshop, pp.

....Eshelman, Caruana and Schaffer (1989) described two recombination biases apparent in traditional non adaptive crossover operators: Positional Bias and Distributional Bias. Positional bias exists when the creation of a new individual is dependent upon the location of the alleles in the chromosome. Booker (1992) showed that, of the nrecombination operators, one point crossover has the highest positional bias. Booker showed that for n L 2 (where n is the number of crossover points and L is the length of the chromosome) the positional bias tends to decrease as n increases for n point recombination. ....

Booker, L.B. (1992) Recombination Distribution for Genetic Algorithms. Proceedings of Foundations of Genetic Algorithms 2, 29-44. Morgan Kauffman.

....over many decades. Several hypotheses and models have been proposed to explain why sexual reproduction is maintained in most organisms in spite of the high cost associated with it [Wil75, MS78, ML88] The GA theory community is beginning to pay attention to results coming from evolutionary biology [Boo93, Lev91, Lev95], and in particular from the field of population genetics. There is much more to be learned that is of potential interest to GA theory. In this paper we reproduce, using GAs and hence finite populations some interesting results obtained with an analytical model using infinite ....

Lashon B. Booker. Recombination distributions for genetic algorithms. In L. Darrell Whitley, editor, Proceedings of the Second Workshop on Foundations of Genetic Algorithms, pages 29--44, San Mateo, July 26-- 29 1993. Morgan Kaufmann.

....the population density through crossovers. In the case of EAs using discrete value chromosomes, Booker theoretically showed the properties of crossovers based on the recombination distribution inspired by Geiringer s result, a kind of the probability at which schemata is generated by crossovers [ Booker, 1992 ] Our analysis in the case of EAs using real number chromosomes is based on the relation between the population densities before and after crossovers, and the change of the mean values and covariances on loci. For this purpose, we have given a framework of the description for the change of the ....

....through the crossovers and the deviations do not change. Although these results are the ones on only the second order moments, they support the result which Qi and Palmieri derived [ Qi and Palmieri, 1994b ] that is, a kind of extension of Geiringer s results in the case of discrete values [ Booker, 1992 ] to the case of real values. The average crossover makes not only the correlations between the different coordinates but also the deviation in the population decrease. In the case of the unfair average crossover, the correlations between the different coordinates in the population change ....

Lashon B. Booker. Recombination Distributions for Genetic Algorithms. In FOGA--92, Proceedings of Workshop on the Foundations of Genetic Algorithms and Classifier Systems, pages 29--44, Morgan Kaufmann, 1992.

....Eshelman, Caruana and Schaffer (1989) described two recombination biases apparent in traditional non adaptive crossover operators: Positional Bias and Distributional Bias. Positional bias exists when the creation of a new individual is dependent upon the location of the alleles in the chromosome. Booker (1992) showed that, of the n point recombination operators, one point crossover has the highest positional bias. Booker showed that for n L 2 (where n is the number of crossover points and L is the length of the chromosome) the positional bias tends to decrease as n increases for n point ....

Booker, L.B. (1992) Recombination Distribution for Genetic Algorithms. Proceedings of Foundations of Genetic Algorithms 2, 29-44. Morgan Kauffman.

....over many decades. Several hypotheses and models have been proposed to explain why sexual reproduction is maintained in most organisms in spite of the high cost associated with it [Wil75, MS78, ML88] The GA theory community is beginning to pay attention to results coming from evolutionary biology [Boo93, Lev91, Lev95], and in particular from the field of population genetics. There is much more to be learned that is of potential interest to GA theory. In this paper we reproduce, using GAs and hence finite populations some interesting results obtained with an analytical model using infinite ....

Lashon B. Booker. Recombination distributions for genetic algorithms. In L. Darrell Whitley, editor, Proceedings of the Second Workshop on Foundations of Genetic Algorithms, pages 29--44, San Mateo, July 26-- 29 1993. Morgan Kaufmann.

....does not use cut points, but simply uses a global parameter to indicate the likelihood that each variable should be exchanged between two parents. Considerable experimental and theoretical work has investigated the differences between these forms of recombination. Spears and De Jong (1992) Booker (1992), and Vose and Liepins (1991) provide theoretical comparisons. Despite the emphasis on recombination within the GA community, interest in mutation has increased recently, partly due to the influence of the ES and EP communities. Schaffer and Eshelman (1991) have experimentally shown that mutation ....

.... For an analysis of deception, see Grefenstette (1992) As we understand better the strengths and weaknesses of our current evolution ary models, it is also important to revisit the biological and evolutionary literature for new insights and inspirations for enhancements. Booker (1992) has recently pointed out the connections with GA recombination theory to the more general theory of population genetics recombination distributions. Mu . hlenbein (1993) has concentrated on EAs that are modeled after breeding practices. In the EP community, Atmar (1992) highlights some errors ....

Booker, L. B. (1992) Recombination distributions for genetic algorithms. Proceedings of the Foundations of Genetic Algorithms Workshop. Vail, CO: Morgan Kaufmann.

....[18] characterized the effects of more general recombination operators in genetic systems Gene 1 Gene 3 Gene 2 Chromosome Gap y3 Gap y2 Gap y1 Linkage = y1 y2 y3 y1 y2 y3 = 1 2 2 2 Figure 3.1: A three gene building block with its inter gene gaps and linkage constraints. and Booker [6] specialized these calculations to commonly used crossover operators in the GA. With some straightforward manipulation, the predictions of their model are seen to be equivalent to those made above. The y i are commonly referred to in probability theory as spacings, and we will use this name for ....

....has five peaks of unequal size unevenly spaced over [0,1] It is defined by f(x) e Gamma2ln2( x Gamma0:1) 0:8) 2 sin 6 (5 (x 3=4 Gamma 0:05) A.4. 6 Function 5 This is the two dimensional function that is a variant of Himmelbau s function [46] Himmelbau s function is defined over [ 6,6] Theta[ 6,6] as f(x,y) x 2 y Gamma 11) 2 (x y 2 Gamma 7) 2 . This is a minimization problem. The function values are multiplied by 1 to convert it into a maximization problem 4 . 4 This is simply an implementation requirement and since tournament selection is used, the ....

[Article contains additional citation context not shown here]

L. B. Booker, "Recombination distributions for genetic algorithms," AI Technical Center, The MITRE Corporation, 1992.

No context found.

L. Booker. Recombination distribution for genetic algorithms. In Foundations of Genetic Algorithms 2, pages 29--44. Morgan Kaufmann, 1993.

No context found.

L. Booker. Recombination distribution for genetic algorithms. In Foundations of Genetic Algorithms 2, pages 29--44. Morgan Kaufmann, 1993.

No context found.

L. B. Booker. Recombination distributions for genetic algorithms. In FOGA-92, Foundations of Genetic Algorithms, Vail, Colorado, 24--29 July 1992. Email: booker@mitre.org.

No context found.

Booker, L. B. (1992) Recombination distributions for genetic algorithms. In FOGA-92, Foundations of Genetic Algorithms, Vail, Colorado.

No context found.

L. B. Booker, "Recombination distributions for genetic algorithms," in FOGA-92, Foundations of Genetic Algorithms, (Vail, Colorado), 24--29 July 1992. Email: booker@mitre.org.

No context found.

Booker, L.B. (1993). Recombination distributions for genetic algorithms. In L.D. Whitley, editor, Foundations of Genetic Algorithms 2, pp. 29--44, San Mateo:MorganKaufman. Bulmer, M.G. (1980). The Mathematical Theory of Quantitative Genetics. Oxford: Clarendon Press.

No context found.

Booker, L.B. (1993). Recombination distributions for genetic algorithms. In L.D. Whitley, editor, Foundations of Genetic Algorithms 2, pp. 29--44, San Mateo:MorganKaufman. Bulmer, M.G. (1980). The Mathematical Theory of Quantitative Genetics. Oxford: Clarendon Press.

No context found.

Booker, L. B. (1992). Recombination distributions for genetic algorithms. In FOGA-92, Foundations of Genetic Algorithms, Vail, Colorado. Email: booker@mitre.org.

No context found.

Booker, L. B. (1992). Recombination distributions for genetic algorithms. In Whitley, D., editor, Foundations of Genetic Algorithms - 2, 29--44. Morgan Kaufmann.

No context found.

L. B. Booker, " Recombination distributions for genetic algorithms ", Foundation of Genetic Algorithms - 2, Morgan Kaufmann Publishers, 1993.

No context found.

Booker, L. B. (1993). Recombination distributions for genetic algorithms. In Whitely D. (Ed.) Foundation of Genetic Algorithms - 2. Morgan Kaufmann Publishers.

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