Z. Michalewicz, Genetics Algorithms + Data Structures = Evolution Programs, WNT, Warsaw 1996  Home/Search   Document Not in Database   Summary   Related Articles   Check

....methods that find the global optimum only with high probability, and methods that guarantee to find a global optimum with some accuracy. An important class belonging to the former type are the stochastic methods [25] A number of techniques such as simulated annealing [25] and genetic algorithms [16] use analogies to physics and biology to approach the global optimum. The most important class of methods of the second type are branch and bound methods [11] They originate from combinatorial optimization, where global optima are also wanted but the variables are discrete and take several values ....

Z. Michalewicz, Genetics Algorithms + Data Structures = Evolution Programs, WNT, Warsaw 1996

....on mass extinction by Fogel et al. 5] was concerned with an application of mass extinction to EP. Relative to clas 1 The term classic , which we use in this paper, refers to simple EAs with basic operators and without population structure. For a survey of these algorithms see for instance [11]. sic EP, the extinction EP was able to outperform on two of eight functions regarding the mean fitnesses. In this paper, we introduce a new mass extinction approach, which is based on the application of self organized criticality (SOC) to a spatial diffusion model (also known as the cellular ....

MICHALEWICZ, Z. Genetic Algorithms + Data Structures = Evolution Programs. Springer, Berlin, 1992.

....for the strategies they use to interact with each other. Evolutionary algorithms employ this powerful design philosophy to find solutions to hard problems. Generally speaking, evolutionary techniques can be viewed either as search methods, or as optimization techniques. As written by Michalewicz [21]: Any abstract task to be accomplished can be thought of as solving Combining Evolutionary and Fuzzy Techniques in Medical Diagnosis 7 a problem, which, in turn, can be perceived as a search through a space of potential solutions. Since usually we are after the best solution, we can view ....

....complexity. They can thus be assimilated within other methods from machine learning, taking advantage of experience gained in this latter domain. In the evolutionary algorithm community there are two major approaches for evolving such rule systems: the Michigan approach and the Pittsburgh approach [21]. A more recent method has been proposed specifically for fuzzy modeling: the iterative rule learning approach [12] These three approaches are briefly 12 Chapter 0 described below. The Michigan approach. Each individual represents a single rule. The fuzzy inference system is represented by the ....

Z. Michalewicz. Genetic Algorithms + Data Structures = Evolution Programs. Springer-Verlag, Heidelberg, third edition, 1996.

....of units were tried. With less than 25 units, results were not satisfactory. With more than 25 units, results show no evidence of improvements, while training times were longer. 5. We use classical Genetic Algorithms and Evolutionary Computation principles such as described in the literature [9, 13]. 6. N represents the number of con ict con gurations on which each element of the population is tested while n represents the number of elements in the population. 7. The GA is not very sensitive to the exact form of the tness function. The one choosen is both simple and ecient. ....

Z. Michalewicz. Genetic algorithms+data structures =evolution programs. Springer-Verlag, 1992. ISBN: 0-387-55387-.

....45 100 1 1000 1 1 1 287569.3725 6 45 100 1 1 0 1 1 16180.3399 7 45 100 1 1 1000 1 1 16180.3399 8 45 100 1 1 1 0.01 1 10000.5000 9 45 100 1 1 1 1 0. 01 431004.0987 10 45 100 1 1 1 1 100 10000.9999 Table 2: Parameter sets for linear quadratic system The ten test cases described by Michalewicz [Mic92] are all implemented in the m file objlinq.m. The values of the parameter sets are shown in table 2. The solution obtained for the first test case is shown in figure 3. u(k) f(x,u) 70 60 50 40 30 20 10 0 10 1 5 9 13 17 21 25 29 33 37 41 45 Fig. 3: Optimal control vector for ....

Michalewicz, Z.: "Genetic Algorithms + Data Structures = Evolution Programs"; Berlin, Heidelberg, New York: Springer, 1992

....of generating candidate solutions from the neighborhood of any particular solution. Those drawbacks can be avoided by using more sophisticated stochastic optimization techniques such as : genetic algorithms. 3.1. 3 Classical Evolutionary Computation or Genetic Algorithm Genetic algorithms (GAs) [8, 9, 5, 11, 10, 4, 14] are problem solving systems based on principles of evolution and heredity. A GA maintains a population of individuals,P (t) x 1 ; x 2 ; xn at iteration t. Each individual represents a potential solution to the problem at hand and is implemented as some (possibly complex) data structure S. ....

....the coding which is one of the success keys of genetic algorithms. 2. Reproduction (Selection) Reproduction is the selection of individuals with respect to the probability distribution based on the the tness values. Fitter individuals have better chances of getting selected for reproduction [11]. Di erent kinds of selection processes can be applied for a speci c problem; one of the most simple is the Roulette Wheel Selection Process : Each chromosome has a certain number of slots proportional to its tness value. The selection process is based on spining the wheel many times; each time ....

Z Michalewicz. Genetic algorithms + Data Structures = Evolution Programs. Springer-verlag, 1992.

....evolution (LIMS) and perform their comparative study. Results showed LIMS has more stable performance than LISS. I. INTRODUCTION Genetic algorithms (GAs) traditionally use bit string representation. However, in recent years many researchers have concentrated on using real valued genes in GAs [Michalewicz 94] Surry 96] Eshelman 97] Ono 99] Theoretical studies of real coded GAs have also been performed [Goldberg 91] Eshelman 97] Qi 94] Kita, 99] Higuchi, 00] Previous studies [Tsutsui 99] Higuchi 00] have proposed simplex crossover (SPX) for real coded GAs. SPX has various good ....

Michalewicz, Z. : Genetic algorithms+data structures=evolution program, Springer-Verlag (1994).

....linkage between the solutions in the phenotype space and the chromosomes in the genotype space, as illustrated in Figure 8.2, and selecting an appropriate encoding scheme is a key issue for GA s. The two most common encoding schemes are binary and real (Holland, 1975; Goldberg, 1989; Davis, 1991; Michalewicz, 1994) In binary coding, each chromosome is a string of binary bits which, with a suitable scaling, describe candidate solutions. A real number coding uses a vector of k real numbers to encode the solution vector (Goldberg, 1990; Janilow and Michalewicz, 1991; Wright, 1991) One advantage of real ....

....from the population average, the fitness is F # k = ck)#. This means that any individual worse than c standard deviations from the population mean (k = c) is not selected at all. The usual value of c reported in the literature is between 1 and 5. Boltzmann Selection. Boltzmann selection (Michalewicz, 1994) is a nonlinear scaling method for proportionate selection, using the following scaling function: F # k = e F k T where T is a user defined control parameter. The selection pressure can be adjusted by assigning T high or low. C.1.2 Rank Based Selection. An alternative way to control the ....

[Article contains additional citation context not shown here]

Michalewicz, Z. (1994). Genetic Algorithms + Data Structures = Evolution Program. Springer-Verlag.

....for the strategies they use to interact with each other. Evolutionary algorithms employ this powerful design philosophy to find solutions to hard problems. Generally speaking, evolutionary techniques can be viewed either as search methods, or as optimization techniques. As written by Michalewicz [10]: Any abstract task to be accomplished can be thought of as solving a problem, which, in turn, can be perceived as a search through a space of potential solutions. Since usually we are after the best solution, we can view this task as an optimization process. Three basic mechanisms drive ....

....learning, taking advantage of experience gained in this latter domain. In the evolutionary algorithm community Fuzzy CoCo: Accuracy and Interpretability by Means of Coevolution 7 there are two major approaches for evolving such rule systems: the Michigan approach and the Pittsburgh approach [10]. A more recent method has been proposed specifically for fuzzy modeling: the iterative rule learning approach [6] These three approaches are briefly described below. The Michigan approach. Each individual represents a single rule. The fuzzy inference system is represented by the entire ....

Z. Michalewicz. Genetic Algorithms + Data Structures = Evolution Programs. Springer-Verlag, Heidelberg, third edition, 1996.

....[9, 4] GAs have to deal with limited population sizes and a limited number of generations. This limitation can lead to premature convergence, which means that the algorithm gets stuck at local optima. A lot of research has been undertaken to overcome premature convergence (for an overview see [10]) Also, experiments have shown that incorporation of problem specific knowledge generally improve GAs. In this paper attention will be paid how to incorporate Air Traffic specific information into a Genetic Algorithm. 5 Application to Airspace Congestion 5.1 Introduction The way this specific ....

Z Michalewicz. Genetic algorithms + Data Structures = Evolution Programs. Springer-verlag, 1992.

....of the rulebased approach include natural representation and ease of integration of learned rules with background knowledge. In the paper we present a new system called EDRL (EDRL, for Evolutionary Decision Rule Learner) which searches for decision rules using an evolutionary algorithm (EA) EAs [11] are stochastic search techniques, which have been inspired by the process of biological evolution. They have been applied to many optimization problems. The success of EAs is attributed to their ability to avoid local optima, which is their main advantage over greedy search methods. Several ....

....others. The results indicated that an entropy based discretization outperformed its competitors, namely equal interval binning, equal frequency binning, and 1R discretizer. 4 Searching for decision rules with EA Our version of evolutionary algorithm follows the general description presented in [11]. In this section we present the following application specific issues: representation, the evolutionary operators, the termination condition and the fitness function. We assume that all continuous valued features have already been discretized. 4.1 Representation Given the class label c k any ....

[Article contains additional citation context not shown here]

Michalewicz, Z.: Genetic Algorithms + Data Structures = Evolution Programs. 3rd edn. Springer-Verlag (1996).

....Dijkstra s shortest path algorithm [1] to address the DBCP. We show that our algorithm solves the DBCP optimally, but may require exponential time in the worst case. From a practical view point, we propose a heuristic algorithm for DBCP which is based on the evolutionary paradigm of computing [7] (See Section 4) We have implemented both the exact algorithm and the heuristic algorithm and tested them on a number of test cases involving large networks. Experimental results are summarized in Section 5. 2 Problem description We now formulate the problem of finding a delay bounded shortest ....

Z. Michalewicz, Genetic Algorithms + Data Structures = Evolution Programs. Springer-Verlag, 1992.

....extremely penalized. As there exists a large number of ways to sort aircraft, the choice of such a filing can be done with genetic algorithms. 4 Genetic Algorithms In this paper, classical Genetic Algorithms and Evolutionary Computation principles such as described in the literature [Gol89, Mic92] are used. The algorithm is used every minutes on the problem defined in section 2.4. 4.1 Data structure During each optimization process, each aircraft trajectory is described by 3 numbers (n, p 0 , t 1 ) n is the number of the path: as detailed in section 2.2, all the alternate path ....

Z Michalewicz. Genetic algorithms + Data Structures = Evolution Programs. Springerverlag, 1992.

....Rule Learner with Multivariate Discretization) combining the two steps: the simultaneous search for threshold values for all continuous valued attributes, which we call multivariate discretization, and the discovery of decision rules. As a search heuristic we use an evolutionary algorithm (EA) [9]. EAs are stochastic techniques, which have been inspired by the process of biological evolution. The success of EAs is attributed to the ability to avoid local optima, which is their main advantage over greedy search methods. Several systems, which employ EAs for learning decision rules (e.g. ....

....ck 1 a copy of a rule selected at random from RS ck 2 , provided that the number of rules in RS ck 1 is lower than maxR . The crossover operator selects at random two rules R c k 1 and R c k 2 from the respective arguments RS ck 1 and RS ck 2 . It then applies an uniform crossover [9] to the strings representing R ck 1 and R ck 2 . 4 Experiments In this section some initial experimental results are presented. We have tested EDRL MD on several datasets from UCI repository [1] Table 1 shows the classification accuracy obtained by our method and C4.5 (Rel. 8) 10] ....

Michalewicz, Z.: Genetic Algorithms + Data Structures = Evolution Programs. 3 rd edn. Springer (1996).

....coecient. Joines and Houck [9] proposed dynamic penalties that vary with the generations. The method of annealing penalties, called Genocop II (for Genetic algorithms for Numerical Optimization of Constrained Problems) is also based on dynamic penalties and was described by Michalewicz and Attia [10, 11]. Adaptive transformation attempt to use the information from the search to adjust the control parameters. This is usually done by examining the tness of feasible and infeasible members in the current population [12] The death penalty method just rejects infeasible individuals. In this method ....

Z. Michalewicz. Genetic algorithms + data structures = evolution programs. Springer Verlag, New York, 3rd edition, 1996.

.... (as a basic structure of an evolutionary algorithm, applicable practically without changes to genetic programming, genetic algorithms, evolution strategies, evolutionary programming, and classif ier systems, may lead to a more complex form like the cultural algorithm, or the genocop algorithm [23]) A simple modifying algorithm is the following cost expression with modif ications (higher order functions) loop, sel, exam, exec, 7 , and user specif ied , init and term, working upon population of programs P (modif ied algorithms) constructed of terminals and built in functions ffi , k , ....

....knowledge of an agent) are invisible, but their cost is not zero. Depending on the cost estimation, global optimization may or may not be successful. 6 Expressing EC, GP, ALife, Agents and DNA Computing 6. 1 Evolutionary Computation and Genetic Programming An arbitrary evolutionary algorithm [23] (which subsumes genetic algorithms, evolutionary programming, evolution strategies, and genetic programming) is very similar to a simple modifying algorithm: ffi (init T F ) initialize population P of chromosomes (eval P ) run P and f ind its cost f itness (loop P ) basic ....

Michalewicz Z., Genetic Algorithms + Data Structures = Evolution Programs, third edition, Springer-Verlag, 1996.

....with Genetic Algorithms Introduction GAs are general purpose global search algorithms that use principles inspired by natural population genetics to evolve solutions to problems. The basic principles of the GAs were rst laid down rigorously by Holland [12] and are well described in many texts as [14]. The basic idea is to maintain a population of knowledge structures that evolves over time through a process of competition and controlled variation. Each structure in the population represents a candidate solution to the speci c problem and has an associated tness to determine which structures ....

Michalewicz Z. (1996) Genetic algorithms + data structures = evolution programs. SpringerVerlag, Berlin/New York, Germany/USA.

....neighborhood of any particular solution. Those drawbacks can be avoided by using more sophisticated stochastic optimization techniques such as : genetic algorithms. f(x) x Local optima Global optimum Figure 4: Global vs. Local Minima 4. 3 Classical Genetic Algorithm Genetic algorithms (GAs) [46, 52, 37, 72, 58, 36, 84] are problem solving systems based on principles of evolution and heredity. A GA maintains a population of individuals,P (t) x 1 ; x 2 ; xn at iteration t. Each individual represents a potential solution to the problem at hand and is implemented as some (possibly complex) data structure S. ....

....the coding which is one of the success keys of genetic algorithms. 2. Reproduction (Selection) Reproduction is the selection of individuals with respect to the probability distribution based on the the fitness values. Fitter individuals have better chances of getting selected for reproduction [72]. Different kinds of selection processes can be applied for a specific problem; one of the most simple is the Roulette Wheel Selection Process : Each chromosome has a certain number of slots proportional to its fitness value. The selection process is based on spinning the wheel many times; each ....

Z Michalewicz. Genetic algorithms + Data Structures = Evolution Programs. Springer-verlag, 1992.

....our nal remarks and propose future extensions for the current project (section 4) 2 Parallel Genetic Algorithm Parallel processing provides an answer to the increasing demand for high performance computing. Taking advantage of the natural parallel characteristics of Genetic Algorithms (GAs) [12], we decided to build a Parallel Genetic Algorithm (PGA) PGAs have successfully solved many global optimization problems [5] Compared to sequential GAs, parallel GAs benet from local selection on each sub population, asynchronous behavior and independence of their processes [14] In [15] for ....

Z. Michalewicz. Genetic algorithms + data structures = evolution programs. Springer-Verlag, 1992.

.... [16, 17, 21] In Germany Rechenberg and Schwefel invented evolution strategies [22, 23] For about 15 years these areas developed separately; it is since the early nineties that they are envisioned as di erent representatives ( dialects ) of one technology that was termed evolutionary computing [1, 2, 3, 10, 20]. It was also in the early nineties that a fourth stream following the general ideas has emerged: Koza s genetic programming [4, 18] The contemporary terminology denotes the whole eld by evolutionary computing, or evolutionary algorithms, and considers evolutionary programming, evolution ....

Z. Michalewicz. Genetic Algorithms + Data structures = Evolution programs. Springer, Berlin, 3rd edition, 1996.

....we propose a new method of learning decision rules, which uses an evolutionary algorithm (EA) EAs are stochastic search techniques, which have been inspired by the process of biological evolution. They have been applied to many search and optimization problems including learning decision rules [2, 6, 8]. There are two key issues in our approach. The first one is the use of two non standard recombination operators, which we call changing condition operator and insertion operator. The second issue is the application of entropy based discretization [5, 3] which allows us to effectively deal with ....

....: c K . Instead we merge all the classes different from c k creating a class c Gamma k . Then we run discretization algorithm and finally we generate decision rules. 3 Searching for decision rules with EA Our version of evolutionary algorithm follows the general description presented in [8]. In this section we describe the following application specific issues: representation, the evolutionary operators, the termination condition and the fitness function. We assume that all continuous valued features have already been discretized. Representation Given the class label c k any ....

[Article contains additional citation context not shown here]

Z. Michalewicz. Genetic Algorithms + Data Structures = Evolution Programs. SpringerVerlag, 3 rd edition, 1996.

.... solution that would not be acceptable at the next shift, the time window is increased of seconds during the optimisation process (see gure 3) 3 GAs applied to AGTO In this paper, classical Genetic Algorithms and Evolutionary Computation principles such as described in the literature [Gol89, Mic92] are used. The algorithm is used every minutes on the problem de ned in section 2.3. 3.1 Data structure During each optimisation process, each aircraft trajectory is described by 3 numbers (n, t 0 , t 1 ) n is the number of the path : as detailed in section 2.2, all the alternate paths ....

Z Michalewicz. Genetic algorithms + Data Structures = Evolution Programs. Springerverlag, 1992.

No context found.

Z Michalewicz, Genetic algorithms + Data Structures = Evolution Programs, Springer-verlag, 1992.

No context found.

Michalewicz, Z. Genetic Algorithms + Data Structures = Evolution Programs. Springer-Verlag, 3rd edition, (1996).

No context found.

Z. Michalewicz, Genetic Algorithms + Data Structures = Evolution Programs. SpringerVerlag, New York, 1992.

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