Eshelman, L. (1991). The CHC adaptive search algorithm: how to have safe search while engaging in nontraditional genetic recombination. In Rawlings, G. J. E., editor, Foundations of genetic algorithms, pages 265--283, San Mateo. Morgan Kaufmann.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....GAs are associated to the use of a binary representation, but nowadays you can find GAs that use other types of representations. A GA usually applies a recombination operator on two solutions, plus a mutation operator that randomly modifies the individual contents to promote diversity. A CHC [7] is a non traditional GA which combines a conservative selection strategy (that always preserves the best individuals found so far) with a highly disruptive recombination (HUX) Certain highly disruptive crossover operator provide more effective search in many problems, which represents the core ....

L. Eshelman. The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In Foundations of Genetic Algorithms, pages 26583. Morgan Kaufmann, 1991.

....[Ackley, 1987] Ackley periodically restarts a search in an attempt to avoid local optima and increase the quality of solutions. Eshelman s CHC algorithm, a genetic algorithm with elitist selection and cataclysmic mutation, also restarts search when the population diversity drops below a threshold [Eshelman, 1991]. Other related work includes Koza s automatically defined functions [Koza, 1993] and Schoenauer s constraint satisfaction method [Schoenauer and Xanthakis, 1993] More recently, Ramsey and Grefenstette come closest to our approach and use previously stored solutions to initialize a genetic ....

....CIGAR over these 50 problems. Both algorithms use exactly the same genetic algorithm parameters and all results reported are averages over 10 runs. We used a population of 100, and ran the algorithms for 100 generations. Eshelman s elitist selection scheme described in Eshelman s paper was used [Eshelman, 1991]. In this scheme, offspring double the size of the population, from N to 2N , and the next generation s population is composed of the best N individuals. Greedy crossover [Grefenstette et al. 1985] with a probability of 1:0 and swap mutation with probability 0:5 that an individual would be ....

Eshelman, L. J. (1991). The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In Rawlins, G. J. E., editor, Foundations of Genetic Algorithms-1, pages 265--283. Morgan Kauffman.

.... these, MOGA s have been introduced successfully in [2] Furthermore, it was shown that the so called best selection helps to find the extreme Pareto solutions [3] This form of selection picks up the best individuals among parents and children for the next generation in a manner similar to CHC [4]. The extreme Pareto solutions are the optimal solutions of the single objectives. By examining the extreme Pareto solutions, the quality of Pareto solutions can be measured. The present MO problem will be solved by using MOGA coupled with the best selection. II. APPROACH In GA s, the natural ....

L. J. Eshelman, "The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination," in Foundations of Genetic Algorithms. San Mateo, CA: Morgan Kaufmann, 1991, pp. 265--283.

....We use the roulette wheel method described in Goldberg to implement this probabilistic strategy [3] We chose the injection interval, the number of generations between injections, to be 2 (N)# where N is the population size. This formula reflects the takeover time when using CHC selection [1]. We inject three cases into the population (10 of the population size) every 2 (30)# = 5 generations. Previous work explains these parameter values [6] All plots are averages over ten (10) runs. CIGAR uses a population of size 100 run for a maximum of 150 generations to solve 4 bit input, ....

L. J. Eshelman. The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In G. J. E. Rawlins, editor, Foundations of Genetic Algorithms-1, pages 265--283. Morgan Kauffman, 1991.

....whenever this occurs in the optimization process. In these algorithms, the optimization is performed in a number of consecutive phases where the best individual survives from one phase to the next. 5.6. 1 CHC algorithm The CHC algorithm suggested by Eshelman combines several ideas in one model [45]. The algorithm uses binary encoding and a number of techniques to both slow down convergence and repair a converged population. The genetic convergence is slowed by the way recombination is performed. First, the CHC algorithm uses a special uniform crossover operator that creates two o#spring by ....

....occurs when the di#erence threshold has dropped to zero and no new o# spring has been accepted for a number of generations. In this case, the population is reinitialized with copies of the best individual where a large part of the genome is mutated. Eshelman suggests to flip 35 of the bits [45]. An interesting aspect of CHC is that it use selection and recombination during the optimization process and only mutation when the population is restarted. The CHC algorithm was compared with a traditional GA on ten numerical benchmark problems, four so called deceptive functions, and a ....

[Article contains additional citation context not shown here]

Eshelman, L. J. (1991). The CHC Adaptive Search Algorithm: How to Have Safe Search When Engaging in Nontraditional Genetic Recombination. In Rawlins, editor, Foundations of Genetic Algorithms, pages 265--283.

....is artificially imposed) these are initially generated in such a way that the hamming distance between them is maximized. Crossover is allowed between individuals belonging to the two distinct populations only. Note that the concept of restricted mating through hamming distance was also used in [10, 11]. The other genetic operators are applied classically. As crossover is allowed between these two dissimilar groups only, a greater degree of diversity is introduced in the population leading to greater exploration in the search. At the same time conventional selection 296 S. BANDYOPADHYAY et al. ....

L. J. Eshelman, The CHC adaptive search algorithm: how to have safe search when. engaging in nontraditional genetic recombination, in: G. J. E. Rawlins (Ed.), Foundations of Genetic Algorithms, Morgan Kaufmann Publishers, Los Altos, CA, 1991, pp. 265-283.

....is used. This strategy has been shown to give good performance on certain test problems. Goldberg initially described this method of implementation for serial GAs [5] The motivation is to maximize the number of schemata sampled during genetic search. Cataclysmic mutation, as described by Eshelman[4] also provides a method of restarting the GA search by reinitializing the population. During reinitialization, the best string in the population at the time of convergence is used as a template by randomly complementing some proportion of the bits; the resulting strings are used to reinitialize ....

Eshelman, L., (1991) "The CHC Adaptive Search Algorithm: How to have Safe Search When Engaging in Nontraditional Genetic Recombination." Foundations of Genetic Algorithms. G. Rawlins, ed. Morgan Kaufmann.

....occurring in the subgraph being repartitioned slightly less costly. New biases are explicitly and partially randomly constructed from the parent(s) for each operation. 3. 3 The CHC adaptive search algorithm The genetic algorithm framework chosen was Eshelman s CHC adaptive search algorithm, [6]. It has been shown to work successfully on a wide range of problems (e.g. 13, 17] with the same parameter settings and, importantly for partitioning large graphs, it uses a small population of 50 individuals. This allowed the simulations to run in a computer s memory. Its main features are: an ....

L. J. Eshelman. The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In G. J. E. Rawlins, editor, Foundations of Genetic Algorithms, pages 265--283. Morgan Kaufmann, San Mateo, 1991.

....the partitions. Each of the 50,000 random trials was generated in the same way. 1000 generations, giving 50,000 evaluations of JOSTLE, were allowed for each run of the genetic algorithm. The genetic algorithm described here is a very simplified instance of the CHC Adaptive Search Algorithm [13] but lacks incest prevention and restarts. The experiments performed showed that the genetic algorithm was able to produce new best individuals until near the completion of the allotted evaluations. 3.4 FITNESS FUNCTION AND IMBALANCE The fitness of a partition (to be maximised) was defined to ....

.... The fact that members of a population are only ever discarded when offspring of greater fitness are generated is known as an elitist strategy [11] It is appropriate in this case because it encourages hill climbing, and because most of the offspring generated are not of very high quality [13] 3.6 RELATED WORK Martin and Otto [16] have also used a hybrid approach to graph partitioning. Their technique applied random changes to a partition, which was then subject to a local optimisation scheme (Kernighan Lin) to improve it. Further changes and local optimisations were applied ....

J. Eshelman. The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination, in G. J. E. Rawlins, editor, Foundations of Genetic Algorithms, pages 205-218. Morgan Kaufmann, 1991,

....loses diversity before sufficient exploration is done to discover the solutions of interest. A number of techniques have been proposed to prevent or slow down premature convergence. These include crowding [DeJong, 1975] sharing [Goldberg and Richardson, 1987] and partial reinitialization [Eshelman, 1991]. In this section, we propose another method to avoid premature convergence, based on maintaining population diversity. The idea is to use a different mutation rate at each string position, and this rate depends on the convergence of the population at that string position. If a string position is ....

Eshelman, L. (1991). The CHC adaptive search algorithm: how to have safe search while engaging in nontraditional genetic recombination. In Rawlings, G. J. E., editor, Foundations of genetic algorithms, pages 265--283, San Mateo. Morgan Kaufmann.

....has yet to be demonstrated in the literature. For this reason, many researchers have turned to other optimization techniques for parallel applications. One search technique that has received a good deal of recent attention with respect to parallel application is the genetic algorithm (GA) [8, 18, 23, 24, 26, 33, 35, 36, 45, 48, 64, 70, 81, 102]. The GA is a general purpose evolutionary search paradigm that operates on a set or population of solutions as opposed to a single solution as in SA, allowing for efficient mapping to multiprocessor environments. As with SA, the GA exhibits elements of both randomization and local search. ....

....while the lowest ranked strings tend not be selected at all. The major drawback to rank based selection is the overhead associated with sorting the population according to fitness. Another improvement to the SGA that has been proposed in the literature is optimization of the control parameters [23, 26, 33, 36]. Recall that the control parameters are comprised of the crossover rate p c , the mutation rate p m , and the size of the population n. The optimal values for these parameters are application dependent [36] but a number of considerations in choosing appropriate parameter values have been ....

[Article contains additional citation context not shown here]

L.J. Eshelman, "The CHC Adaptive Search Algorithm: How to Have Safe Search When Engaging in Nontraditional Genetic Recombination," in G. Rawlins, ed., Foundations of Genetic Algorithms, Morgan Kaufmann, San Mateo, CA, 265-283, 1991.

....valid for nite populations was given by Rogers and Pr ugel Bennett [5, 6] This Complex Systems, 11 (1997) 1 1 ; c 1997 Complex Systems Publications, Inc. 2 Adam Pr ugel Bennett allowed comparison of di erent selection strategies, such as, steadystate, generation gap [7] CHC models [8], evolutionary strategies and stochastic universal sampling [9] However, the approximation, although physically plausible, is not systematic. That is, there are no bounds on the errors, nor can we obtain higher order corrections. In this paper we calculate the e ect of ranking (or tournament) ....

L. Eshelman. The chc adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In G. Rawlins, editor, Foundations of Genetic Algorithms 1, pages 265{ 283, San Mateo, 1991. Morgan Kaufmann.

....generational selection G = 1 and for steady state selection G = 1=P . We follow this generalisation and calculate the change in variance for any value of generation gap. The formalism can also be extended to other non traditional selection schemes such as that used in Eshelman s CHC algorithm [12]. Here we confirm analytically an observation made by Schaffer et al. 2] that shows using a numerical Markov chain analysis that a simple model of CHC style selection exhibits half the rate of genetic drift of the traditional genetic algorithm. The simple model of the CHC algorithm is equivalent ....

L. Eshelman, "The CHC Adaptive Search Algorithm: How to Have Safe Search When Engaging in Nontraditional Genetic Recombination," in Foundations of Genetic Algorithms 1, G. Rawlins, Ed., San Mateo, 1991, pp. 265--283, Morgan Kaufmann.

....of the search space, the possibility that a recombination operator generates offspring around the optimum point decreases because a portion of the feasible offspring space located beyond the boundary of the search space is cut away. To see this bias, we provide an analysis using BLX a operator [Eshelman 91] Other recombination operators for real coded GAs such as UNDX [Ono 97] multi parent recombination operators [Tsutsui 98] SPX [Tsutsui 99] also have this kind of bias. An analysis of sampling bias was done from a different angle in [Eshelman 97] Kita 99] Here, for simplicity, without loss ....

....to (max min ) 2 and p m is fixed to 0.2 n for all experiments where min and max are the lower and upper limits of the parameter range on the i th dimension of the search space. 3) Basic evolutionary model The basic evolutionary model we used in these experiments is similar to that of the CHC [Eshelman 91] and Journal of Information Science (in press) l) ES [Schewefel 95] Let the population size be N, and let it, at time t, be represented by P(t) The population P(t 1) is produced as follows: A collection of N 2 pairs is randomly composed, and crossover is then applied to each pair, generating ....

Eshelman, L. J.: The CHC adaptive search algorithm: how to have safe search when engaging in nontraditional genetic recombination, Foundations of Genetic Algorithms, Morgan Kaufmann, pp.265-283 (1991).

....in natural systems, and the intention is to extract relevant features in the mating parents and then, hopefully, produce better tted o spring. The central idea in crossover is to nd the balance point between the preservation of schemata and the e ective recombination as mentioned by Eshelman [Esh91]. In other words, there is a tradeo between the extent to which the good solutions at a particular stage are preserved and the extent to which new sections of the search space are explored. This is to say, that the e ectiveness in GAs depends upon nding the appropriate combination between ....

L.J. Eshelman. The CHC adaptive search algorithm: How to have safe search when angaging in nontraditional genetic algorithm. In G. Rawlins, editor, Foundations of Genetic Algorithms, pages 265-283, 1991.

.... The fact that members of a population are only ever discarded when offspring of greater fitness are generated is known as an elitist strategy, 8] It is appropriate in this case because it encourages hillclimbing, and because most of the offspring generated are not of very high quality, [4]. The random initial population was generated by (for each individual) assigning to every vertex bias values chosen randomly and uniformly from [0, 0.1] and then using JOSTLE to generate a partition. 1000 generations were allowed for each run of the genetic algorithm, giving 50,000 evaluations ....

....randomly and uniformly from [0, 0.1] and then using JOSTLE to generate a partition. 1000 generations were allowed for each run of the genetic algorithm, giving 50,000 evaluations of JOSTLE. The genetic algorithm described here is a very simplified instance of the CHC Adaptive Search Algorithm, [4], but lacks incest prevention and restarts. The experiments performed showed that the genetic algorithm was able to produce new best individuals until near the completion of the allotted evaluations. 3.6 Related work There are a limited number of papers about topics related to the work presented ....

L. J. Eshelman. The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In G. J. E. Rawlins, editor, Foundations of Genetic Algorithms, pages 265--283, San Mateo, 1991. Morgan Kaufmann.

.... tournament selection [Gol89] and ( selection [BHmS91, Rec94, Sch95] triplecompetition (rat race selection) vK97b] and and elitist recombination [TG94] Furthermore we are going to discuss the Breeder genetic algorithm by Muhlenbein et al. MSV94] and the CHC algorithm by Eshelman [Esh91] A selection scheme is assumed to select the parent pairs for generation G i 1 from the individuals in generation G i . In case of a canonical genetic algorithm the parents taken from generation G i are discarded and a fitness proportional selection is applied to their offspring. When using an ....

....2 ; fi) fi x 1 (1 Gamma fi) x 2 Here the first term represents the density of the parents while the second term represented the contribution of the offspring. The vector y is again computed by the formula: y i = X j;k T o (i j; k)x j x k The CHC algorithm is described by Eshelman [Esh91]. The CHC algorithm uses an unbiased selection of parents. Given a parents population of size N , a set of N offspring is produced. The next parent generation is obtained by selecting the N best amongst the N parents and their N offspring. This selection scheme corresponds to ( selection, ....

L.J. Eshelman. The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In G. Rawlins, editor, Foundations of Genetic Algorithms - 1, pages 265--283. Morgan Kaufmann, 1991.

....will evolve over time. An infinite number of algorithms will fit into the generic template just given, but most GAs can be represented by a simple genetic algorithm (see Figure 1.1) While many GAs are based on this simple GA, some are not. Some non traditional genetic algorithms include CHC [13] and GENITOR [50] The CHC algorithm merges the old population with the new, keeping the n most fit strings in the merged population. GENITOR replaces the least fit string of the population with a newly generated string iff the new string has a higher fitness than the old. GIGA and NQ GIGA, two of ....

Eshelman, Larry J. (1991). "The CHC Adaptive Search Algorithm: How to Have Safe Search When Engaging in Nontraditional Genetic Recombination." In Foundations of Genetic Algorithms, Gregory J.E. Rawlins, editor, pp. 265283. San Mateo, CA: Morgan Kaufmann.

....parents simply with N children. The second one is the elitist recombination (ER) model that selects two best individuals among two parents and their two offsprings. The final model is the so called best N (BN) model that selects the best N individuals among N parents and N children similar to CHC [7]. The population size was kept to 100 in all test cases. 3. COMPARISON OF NICHING AND ELITIST MODELS From the techniques described above, five optimization results are shown here for the first test case. Figures 2 to 4 show the results obtained from the simple generational model with the ....

Eshelman, L. J.: The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination, Foundations of Genetic Algorithms, Morgan Kaufmann Publishers, Inc., San Mateo, 1991, pp. 265-283.

....have been found. For their part, works around enhancing the exploratory phase have mostly focussed either on avoiding all chromosomes to rush towards one optimum (multi niched GAs) or on iteratively letting the GA converge and trying to make it converge towards different points each time [21] [12], 24] 4] The first path has been treated for a long time by preventing too similar individuals to recombine (incest prevention) or to avoid too much individuals of being similar (sharing techniques) The interested reader is referred to [37] which have recently reviewed these techniques. ....

Larry J. Eshelman. The CHC adaptive search algorithm: how to have safe search when engaging in nontraditional genetic recombination. In [32], pages 265--283, 1991.

....1987] Ackley periodically restarts a search in an attempt to avoid local optima and increase the quality of solutions. Eshelman s CHC algorithm, a genetic algorithm with elitist selection and cataclysmic mutation, also restarts search when the population diversity drops below a threshold [Eshelman, 1991]. Both approaches only attack a single problem not a related set of problems. Ramsey and Grefenstette come closest to our approach and use previously stored solutions to initialize the genetic algorithm s initial population and thus increase a genetic algorithm s performance in an environment that ....

Eshelman, L. J. (1991). The chc adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In Rawlins, G. J. E., editor, Foundations of Genetic Algorithms-1, pages 265--283. Morgan Kauffman.

....algorithm is applied to the produced o spring before they are added to the population. A new generation is formed by selecting the best individuals from the extended population for survival. A unique feature of our memetic algorithm is the restart mechanism borrowed from Eshelman s CHC algorithm [24]. If the search has converged, e.g. the chances for generating new solutions with better tness have approached zero, the population is diversi ed by mutating all but the best individual. 9 procedure MA; begin initialize population P ; foreach individual i 2 P do i : Local Search(i) ....

....complexity of the local search does not allow evolving much larger populations in reasonable time. Such a small population size leads to a premature convergence of the algorithm, especially in the absence of mutation. To overcome this drawback, the restart technique proposed by Eshelman [24] is employed. During the run, it is checked whether the average distance of the population has dropped below a threshold d = 10, or the average tness of the population did not change for more than 30 generations. If one of these conditions hold, the search is assumed to be converged and the whole ....

L. Eshelman, \The CHC Adaptive Search Algorithm: How to Have Safe Search When Engaging in Nontraditional Genetic Recombination, " in Foundations of Genetic Algorithms, (G. J. E. Rawlings, ed.), pp. 265-283, Morgan Kaufmann, 1991.

....considered a dynamic parameter encoding scheme in which the interval represented by a given gray coded binary gene shrinks over time, triggered based on measures of population convergence. Both pairs of researchers reported good results on a test suite of problems. In his CHC genetic algorithm, Eshelman (1990) employed mass mutation to restart the search when the population had converged. In normal operation CHC used no mutation. During a restart, copies of the best individual found to date were mutated at a high rate (e.g. 35 ) to produce the new population. Typically the best individual was also ....

....for countering convergence and thus encouraging diversity. These include exploitation of population structure and speciation (surveyed in Radcliffe Surry, 1994d) co evolutionary models (e.g. Hillis, 1991; Husbands Mill, 1991) adaptive mutation (Whitley Hanson, 1989) incest prevention (Eshelman, 1990), crowding (De Jong, 1975) and sharing (Goldberg, 1989c) as well as others. Due to the diverse nature of the problems studied here (section 10.4) we used only simple measures based on the selection and replacement regime. Our algorithms enforced uniqueness within the population and used ....

L. J. Eshelman, 1990. The CHC adaptive search algorithm: How to have safe search when engaging in non-traditional recombination. In Foundations of Genetic Algorithms. Morgan Kaufmann (San Mateo).

....distance. The idea is to allow two individuals to reproduce only if they are very similar (i.e. if their phenotypic distance is less than a factor called oe share ) This is intended to produce distinct species (mating groups) in the population [Mitchell 1996] Other researchers such as Eshelman [1991] and Schaffer [1991] did exactly the opposite: they didn t allow mating between individuals that were too similar (they said to be preventing incest ) Smith, Forrest and Perelson [1993] proposed an approach, modelled after the immune system, that can maintain the diversity of the population ....

Eshelman, L. J. 1991. The CHC Adaptive Search Algorithm: How to Have Safe Search when Engaging in Nontraditional Genetic Recombination. In G. E. Rawlins Ed., Foundations of Genetic Algorithms , pp. 265--283. San Mateo, California: Morgan Kaufmann Publishers.

..... 155 A.1 String Representation of Parse Trees . 176 A.2 Experimental Results . 179 xi Abbreviations ANOVA analysis of variance ARG attributed relational graph CHC genetic algorithm variant (Eshelman 1991) COST labelling criterion CROSS crossover type GA genetic algorithm GEN genetic algorithm iteration limit GENITOR genetic algorithm variant (Whitley 1989) HUX half uniform crossover LINES number of lines in a drawing MAP maximum a posteriori probability NODES number of nodes in a graph POP genetic ....

.... GENITOR, which combines rank based selection with the steady state algorithm (Whitley 1989) Ackley s Iterated Genetic Search (Ackley 1987) and Eshelman s CHC algorithm, in which recombination can only occur between chromosomes when the Hamming distance between them is above a certain threshold (Eshelman 1991). An important class of genetic algorithm is the hybrid genetic algorithm, sometimes called memetic algorithm , in which evolutionary optimisation is coupled with a local search step (Davis 1991; Cross et al. 1997; Whitley et al. 1995) or even simulated annealing (Yip and Pao 1995) It is ....

[Article contains additional citation context not shown here]

L. J. Eshelman (1991). The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In G. J. E. Rawlins (Ed.), Foundations of Genetic Algorithms, Volume 1, pp. 265--283. Morgan Kaufmann.

.... as done, for example, in the ( evolution strategy [33] Additionally, we ensure that every solution is contained only once in the population (duplicate checking) Diversification restarts: In our approach, a diversification strategy is borrowed from the CHC algorithm proposed by Eshelman [6]. If the algorithm is said to have converged (the average hamming distance of the population has dropped below a threshold d=10 or there was no change in the population for more than 30 generations) the whole population is mutated except the best individual and the search is restarted. This ....

....BQP, we use the 1 opt local search algorithm as described below and a variant of the uniform crossover for recombination. This variant, called HUX, is biased to create offspring that are equally far away from their parents in terms of the hamming distance, similar to the HUX operator described in [6]. The local search applied to the resulting offspring after recombination is restricted to a region of the search space defined by the two parents: the genes with equal values in the two parents are not modified during local search. Local Search for the BQP The simplest form of local search for ....

L. Eshelman, "The CHC Adaptive Search Algorithm: How to Have Safe Search When Engaging in Nontraditional Genetic Recombination," in Foundations of Genetic Algorithms, (G. J. E. Rawlings, ed.), pp. 265--283, Morgan Kaufmann, 1991.

....to try to find an other optimum 2. the exploitation of the already found basin of attraction to find as efficiently as possible the optimal point in the basin The first point may be solved by restarting the EA with a new population hoping that the EA will not fall into the same basin. Eshelman [Esh91] reports enhanced results using 3 we leave aside niching techniques that aim at avoiding this uniformization and permit an EA to find, and maintain, multiple optima during the same evolution. 1020 1040 1060 1080 1100 1120 1140 1160 1180 1200 0 250 500 750 1000 1250 1500 1750 2000 Generations ....

Larry J. Eshelman. The CHC adaptive search algorithm: how to have safe search when engaging in nontraditional genetic recombination. In [Raw91], pages 265--283, 1991.

....same answer. The reason is that the population at the moment of finding the best result could have been recombined and improved several times, being quite different of the random initial population of a simple GA. This procedure has some resemblance with Eshelman s CHC Adaptive Search Algorithm [14], but in our case we do not use any re feeding of the population through high mutation values when it has stabilized, nor a highly disruptive recombinator operator that produces offspring that are maximally different from both parents. Our approach uses a conventional two point crossover and it ....

Larry J. Eshelman. The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In Gregory E. Rawlins, editor, Foundations of Genetic Algorithms, pages 265--283. Morgan Kaufmann Publishers, San Mateo, California, 1991.

....GA community like Goldberg and Grefenstette [58] 60] 61] Jog, Van Gucht and coworkers [75] 76] 128] H. Muhlenbein and M. Gorges Schleuter [59] D. Whitley and coworkers [133] 134] 126] and references therein) and others [15] 88] 106] 122] 13] 118] 41] 42] 43] 1] 2] 62] 92] [34] [46] 74] Among them, some researchers started to depart from the application of traditional genetic algorithms and started to introduce periods of local search in their strategies for the TSP. Some of them defined this strategy as knowledge augmented methods like Grefenstette and co workers ....

L.J. Eshelman, The CHC Adaptive Search Algorithm: How to Have Safe Search when Engaging in Nontraditional Genetic Recombination, in: Rawlins, Gregory J.E. (ed.) Foundations of Genetic Algorithms (Morgan Kaufmann, San Mateo CA, 1991) 265-283.

....was the first steady state genetic algorithm. Genitor selects two parent individuals by ranking selection and applies mixing to them to produce one offspring, which replaces the worst element of the population. Genitor fits the framework of subsection 4.2.1 with k = 1. Eshelman s CHC algorithm [Eshelman, 1991] is a generational GA where the best individuals are drawn from the combined parent and offspring population to obtain the next generation. This selection strategy is called the ( Evolution Strategy. Duplicates are removed from the population. Then parents for crossover are chosen randomly ....

Eshelman, L. (1991). The chc adaptive search algorithm: how to have safe search while engaging in nontraditional genetic recombination. In Rawlings, G. J. E., editor, Foundations of genetic algorithms, pages 265--283, San Mateo. Morgan Kaufmann.

....which accepts non improvement moves . Although Ref. 7] is referenced as an example of such a method (those which accept non improvement moves at the local optimization phases) in that paper the binary perceptron learning problem is first addressed with a simple descent method (see also Ref. [8] as well as the papers of Gorges Schleuter, Muhlenbein, and others cited therein) Regardless these details, the techniques have many analogies with new hybrid genetic algorithms as well as former methods like Scatter Search introduced by F. Glover in 1977 (see Refs. 40] 45] in [7] and Refs. ....

L.J. Eshelman. The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In G.J.E. Rawlins, editor, Foundations of Genetic Algorithms, pages 265--283, San Mateo, CA, USA, 1991. Morgan Kaufmann Publishers.

....the population size of a memetic algorithm is typically small compared to genetic algorithms: a size of 10 up to 40 is common in the MA. Such a small population size may lead to a premature convergence of the algorithm. To overcome this drawback, the restart technique has been proposed by Eshelman [13]. In our MA we have adapted this technique to achieve a much more robust search algorithm: Upon convergence (the average distance of the population has dropped below a threshold d=10 or there was no change in the population for more than 30 generations) the whole population is mutated except the ....

L. Eshelman, "The CHC Adaptive Search Algorithm: How to Have Safe Search When Engaging in Nontraditional Genetic Recombination," in Foundations of Genetic Algorithms, (G. J. E. Rawlings, ed.), pp. 265--283, Morgan Kaufmann, 1991.

.... Theta 5 units with a rectangular hole centered at (0; 0) that have sides of 2 Theta 1 units. The initial population consists of 40 random flaw configurations. The initial flaws are of rectangular shape with randomly generated sides and center locations. We use the CHC elitist selection strategy [3] which always preserves the best individual, a 0:85 probability of crossover, and a 0:055 mutation rate on both problems in this section. Y X o 2.0 1.0 5.0 10.0 Figure 3: Original One Hole Problem The terminal condition in the proposed method requires that the average error, computed as the mean ....

Larry J. Eshelman. The chc adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In Gregory J. E. Rawlins, editor, Foundations of Genetic Algorithms-1, pages 265--283. Morgan Kauffman, 1991.

....(Ackley 1987) Ackley periodically restarts a search in an attempt to avoid local optima and increase the quality of solutions. Eshelman s CHC algorithm, a genetic algorithm with elitist selection and cataclysmic mutation, also restarts search when the population diversity drops below a threshold (Eshelman 1991). Other related work includes Koza s automatically defined functions (Koza 1993) and Schoenauer s constraint satisfaction method (Schoenauer Xanthakis 1993) More recently, Ramsey and Grefenstette come closest to our approach and use previously stored solutions to initialize a genetic ....

Eshelman, L. J. 1991. The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In Rawlins, G.

....tuned by a tness biased selection of the parents. An alternative is to tune the selective pressure by a tness biased selection of the individuals that should be replaced. Examples of GAs with a replacement strategy are Genitor (Whitley,1989) Syswerda s GA (Syswerda, 1991) Eshelman s CHC (Eshelman, 1991) and ( B ack, 1991) The rst two are steady state implementations while the latter two are generational. To illustrate the modeling of selection intensity for replacement strategies we consider a rather general and easy to tune selection scheme that combines the tness biased parent ....

Eshelman L. (1991). The CHC Adaptive Search Algorithm: How to have safe search when engaging in nontraditional genetic recombination. Foundations of Genetic Algorithms I ed. G.Rawlins. Morgan Kaufmann.

....We use elitist selection and randomly choose individuals to mate. Each pair of parents produces two children through crossover and or mutation. After sorting the new individuals together with the old individuals, we choose the best N (population size) individuals to be in the new population (Eshelman, 1990). The velocity estimation problem has a very large search space. If we use five bits to encode the velocity range from 1.8 to 7.2 km sec, the total number of possible models is 2 5 ThetaN x ThetaN z . Since N x and N z are 40 and 8or9 in our experiments (and can be much larger) leading to a ....

Eshelman, L. J. (1990). The chc adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In Rawlins, G. J. E., editor, Proceedings of the Foundations of Genetic Algorithms Workshop - 1, San Mateo, CA. Morgan Kauffman.

....0 swap sites Figure 4: Swap mutation 3.4 Selection When using traditional roulette wheel selection, the best individual has the highest probability of survival but does not necessarily survive. We use CHC selection to guarantee that the best individual will always survive in the next generation [Eshelman 91] In CHC selection if the population size is N , we generate N children by using roulette wheel selection, then combine the N parents with the N children, sort these 2N individuals according to their fitness value and choose the best N individuals to propagate to the next generation. From Figure ....

L. J. Eshelman. "The chc adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination." In G. J. E. Rawlins, editor, Foundations of Genetic Algorithms-1, pages 265--283. Morgan Kauffman, 1991.

No context found.

Eshelman, L. (1991). The CHC adaptive search algorithm: how to have safe search while engaging in nontraditional genetic recombination. In Rawlings, G. J. E., editor, Foundations of genetic algorithms, pages 265--283, San Mateo. Morgan Kaufmann.

No context found.

Eshelman, L. (1991). The CHC adaptive search algorithm: how to have a safe search when engaging a non-traditional genetic recombination. In: Rawlings, G. (ed.) Foundations of Genetic Algorithms (FOGA-1), Morgan Kaufman, San Francisco, USA. pp. 265-283.

No context found.

L.J. Eshelman. The CHC Adaptive Search Algorithm: How to Have Safe Search when Engaging in Nontraditional Genetic Recombination. In G. Rawlins, editor, Foundations of Genetic Algorithms, pages 265--283, San Mateo, 1991. MorganKaufman.

No context found.

L. Eshelman, "The CHC Adaptive Search Algorithm: How to Have Safe Search When Engaging in Nontraditional Genetic Recombination," in Foundations of Genetic Algorithms, (G. J. E. Rawlings, ed.), pp. 265--283, Morgan Kaufmann, 1991.

No context found.

L. Eshelman, \The CHC Adaptive Search Algorithm: How to Have Safe Search When Engaging in Nontraditional Genetic Recombination, " in Foundations of Genetic Algorithms, edited by G. J. E. Rawlings, 265-283, (Morgan Kaufmann, 1991).

No context found.

Larry J. Eshelman. The CHC adaptive search algorithm: how to have safe search when engaging in nontraditional genetic recombination. In [32], pages 265--283, 1991.

No context found.

L. Eshelman, \The CHC Adaptive Search Algorithm: How to Have Safe Search When Engaging in Nontraditional Genetic Recombination," in Foundations of Genetic Algorithms, (G. J. E. Rawlings, ed.), pp. 265-283, Morgan Kaufmann, 1991.

No context found.

CAL. Eshelman, "The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination," in Proceedings of the Foundations of Genetic Algorithms Workshop -1,G. Rawlins, Ed. San Mateo: Morgan Kaufmann, 1991.

No context found.

L.J. Eshelman. The CHC Adaptive Search Algorithm: How to Have Safe Search when Engaging in Nontraditional Genetic Recombination. In G. Rawlins, editor, Foundations of Genetic Algorithms, pages 265--283, San Mateo, 1991. MorganKaufman.

No context found.

Larry J. Eshelman. The CHC Adaptive Search Algorithm: How to Have Safe Search when Engaging in Nontraditional Genetic Recombination. In Gregory E. Rawlins, editor, Foundations of Genetic Algorithms, pages 265--283. Morgan Kaufmann Publishers, San Mateo, California, 1991.

No context found.

L. Eshelman. The CHC adaptive search algorithm: How to have safe search when engaging in nontraditional genetic recombination. In Rawlins, editor, Foundations of Genetic Algorithms, pages 265--283. Morgan Kaufmann Publishers, 1991.

No context found.

Eshelman, L. J.: The CHC adaptive search algorithm: how to have safe search when engaging in nontraditional genetic recombination, Foundations of Genetic Algorithms, Morgan Kaufmann, pp.265-283 (1991).

No context found.

Eshelman, L. (1990). The CHC Adaptive Search Algorithm: How to have safe search when engaging in nontraditional genetic recombination. Foundations of Genetic Algorithms, Bloomington, IN..

First 50 documents  Next 50

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