J. Schaffer, D. Whitley, and L. Eshelman. Combinations of genetic algorithms and neural networks: A survey of the state of the art. In Proceedings of the International Workshop on Combinations of Genetic Algorithms and Neural Networks, pages 1--37, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....networks (ANN s) in which evolution is another fundamental form of adaptation in addition to learning. Evolution can be introduced at various levels of ANN s. It can be used to evolve weights, architectures, and learning parameters and rules. EANN s have been studied widely in recent years [4] [6]. They provide not only an automatic method to design ANN s, but also an approach to study evolution and learning in the same framework [4] 6] This paper is mainly concerned with the evolution of ANN s architectures and weights (including biases) where an evolutionary algorithm is used to ....

....levels of ANN s. It can be used to evolve weights, architectures, and learning parameters and rules. EANN s have been studied widely in recent years [4] 6] They provide not only an automatic method to design ANN s, but also an approach to study evolution and learning in the same framework [4] [6]. This paper is mainly concerned with the evolution of ANN s architectures and weights (including biases) where an evolutionary algorithm is used to evolve ANN s architectures and or weights. Although there have been many studies on how to evolve ANN s more effectively and efficiently [4] 6] ....

[Article contains additional citation context not shown here]

J. D. Schaffer, D. Whitley, and L. J. Eshelman, "Combinations of genetic algorithms and neural networks: A survey of the state of the art," in Proc. Int. Workshop Combinations Genetic Algorithms Neural Networks (COGANN-92), D. Whitley and J. D. Schaffer, Eds. Los Alamitos, CA: IEEE Comput. Soc. Press, 1992, pp. 1--37.

....to evolving ANN architectures. One is the evolution of pure architectures (i.e. architectures without weights) Connection weights will be trained after a near optimal architecture has been found. The other is the simultaneous evolution of both architectures and weights. Schaffer et al. [17] and Yao [18] 21] have provided a comprehensive review on various aspects of evolutionary artificial neural networks (EANN s) A. The Evolution of Pure Architectures One major issue in evolving pure architectures is to decide how much information about an architecture should be encoded into a ....

....explained why it is important in terms of accurate fitness evaluation. The simultaneous evolution of both architectures and weights can be summarized by Fig. 2. The evolution of ANN architectures in general suffers from the permutation problem [27] 28] or called competing conventions problem [17]. It is caused by the many to one mapping from genotypes to phenotypes since two ANN s which order their hidden nodes differently may have different Fig. 1. A typical cycle of the evolution of architectures. Fig. 2. A typical cycle of the evolution of both architectures and weights. The word ....

J. D. Schaffer, D. Whitley, and L. J. Eshelman, "Combinations of genetic algorithms and neural networks: A survey of the state of the art," in Proc. Int. Wkshp. Combinations Genetic Algorithms Neural Networks (COGANN-92), D. Whitley and J. D. Schaffer, Eds. Los Alamitos, CA: IEEE Computer Soc. Press, 1992, pp. 1--37.

....the modeling of the German stock index DAX is attempted with a neuro fuzzy approach. 2. 2 Neural networks and Evolutionary algorithms Fundamental implementations of neural networks generated and tuned by genetic algorithms can be found in a series of publications [41] 42] 43] 44] 45] [46]. The idea behind the implementation of such a hybrid system is the adoption of an evolutionary algorithm for the determination of neural network s weights or the neural network s architecture, or both. In the first case, neural networks are tuned by evolutionary algorithm, rather than generated ....

Schaffer J. D., Whitley D. and Eshelman L. J., Combinations of genetic algorithms and neural networks: a survey of the state of the art, Proc. Int. Workshop on Combinations of Genetic Algorithms and Neural Networks, COGANN'92, pp 1-37, 1992

....we make use of the relation between NN topologies and graphs. Exploiting methods from graph theory we demonstrate a general way to deal with isomorphic structures. The presented approach applied to NN optimization can be regarded as a solution to the so called competing conventions problem [19]. Further, it can be used to save fitness evaluations when used in combination with a graph database. Keywords structure and topology optimization, recurrent neural networks, isomorphism, graph representation, competing conventions problem, rare and frequent structures 1 Introduction In this ....

....reflect deeper insight into the theory of neural networks but demonstrates that there is no systematic and reliable way to generate problem specific network topologies. Results from structure optimization of NNs show that evolutionary algorithms are very suitable to optimize the topology of NNs [2, 19, 21, 27]. In order to tackle the structure optimization problem in the framework of EAs one has to define a genotype space which constitutes the search space, and search operators e.g. mutation or crossover. Each genotype in the search space can be mapped to a NN (simple example: connection matrix as ....

[Article contains additional citation context not shown here]

J. D. Schaffer, D. Whitley, and L. J. Eshelman. Combinations of genetic algorithms and neural networks: A survey of the state of the art. In L. D. Whitly and J. D. SchaffZr, editors, COGANN-92: International Workshop on Combinations of Genetic Algorithms and Neural Networks, pages 1-37, 1992.

.... rule s condition and action are represented by a multi layered perceptron (MLP) The weights of each neural rule being evolved under the actions of the genetic algorithm (GA) 6] The approach is also related to the use of evolutionary computing techniques to produce neural networks in general (see [12] for an early review) In contrast to that work, an LCS based approach is co evolutionary, the aim being to develop a number of (small) cooperative neural networks to solve the given task, as opposed to the evolution of one (large) network. SANE [11] is most similar to the work described here, ....

Schaffer, J., Whitley, D. & Eshelman, L.J. (1992) Combinations of Genetic Algorithms and Neural Networks: a Survey of the State of the Art. In Proceedings of the Conference on combinations of Genetic Algorithms and Neural Networks. IEEE, pp1-37.

....and repeated sampling can lead to some natural insensitivity to noisy feedback. The genetic algorithm based heuristics are highly suited for application to large instances of problems that are hard to model and for which no satisfactory tailored algorithms are available. Readers are referred toSchaffer et al. 1992), Goldberg (1989) and Holland (1992) for more information about genetic algorithms. For large MIP formulations, stochastic global optimization methods such as simulated annealing (SA) and genetic algorithms have recently shown great promise. The SA based approaches operate on one intermediate ....

....value in terms of material handling cost) is selected as the best solution for the problem. Though genetic algorithms are effective search techniques, they are known to be sensitive to control parameters, e.g. population size, rates of crossover (CR) and mutation (MR) methods of selection,etc. (Schaffer et al. 1992). Previous attempts at establishing an analytical measure for optimal choices of control parameters in genetic algorithms have not resulted in any closed form equations due to the complex nature of the interactions, the problem specific nature of the search procedure parameters, and ....

Schaffer, J.D., Whitley, D., and Eshelman, L.J., 1992, Combinations of genetic algorithms and neural networks: a survey of the state of the art,Proceedings of the International Conference on the Combinations of Genetic Algorithms, San Mateo, CA: Morgan Kaufmann Publishers.

....on Evolutionary Computation (CEC 99) Washington, July 6 9, 1999, IEEE Press. 1 1 Introduction The combined application of neural network techniques and evolutionary algorithms turned out to be a very effective tool for solving an interesting class of problems (for a review see e.g. 1] [8], 11] Especially, in situations where a task involves dynamical features like generation of temporal sequences, recognition, storage and reproduction of temporal patterns, or for control problems requiring memory to compute derivatives or integrals, other learning strategies are in general not ....

....of neurons and the architecture of a network. In fact, it develops network topology and parameters like weights and bias terms simultaneously on the basis of a stochastic process. In contrast to genetic algorithms, which are often only used for optimizing a specific feedforward architecture [8], 11] it does not quantize the network parameters like weights and bias terms. With respect to algorithms like, for instance, EPNet [12] it does not include an individual learning procedure, which exists naturally only for feedforward networks and problems where an error function or ....

Schaffer, J. D., Whitley, D., and Eshelman, L. J. (1992), Combination of genetic algorithms and neural networks: A survey of the state of the art, in: Proceedings International Workshop on Combinations of Genetic Algorithms and Neural Networks (COGANN-92), Los Alamitos, CA, IEEE Computer Society Press.

....which is usually a binary or integer string. For continuous (non discrete) parameter problems, genetic algorithms discretizing the search space get moderate performances in comparison with other algorithms like gradient descent algorithms (when the function to optimize is differentiable) [11].Optimizing continuous (non discrete) parameter functions with EAs requires real parameter genetic coding and highly sophisticated algorithms. Some EAs attempt to solve these problems by working directly on continuous data ( 1] 8] 9] The use of EAs has shown to be efficient for boolean ....

J.D. Schaffer, D. Withley, and L.J. Eshelman. Combinations of Genetic Algorithms and Neural Networks: A Survey of the State. In COGANN92. International Workshop on Combinations of Genetic Algorithms and Neural Networks, 1992.

....structure or set of structures. In fact, people have tried optimizing most traditional machine learning structures as well as some nontraditiona l structures using GAs. These structures have ranged from neural network weights and topologies (Gruau Whitley, 1993; Whitley et al. 1990, 1991, 1993; Whitley Schaffer, 1992), to LISP programs (Koza, 1992) to regions of the instance space similar to decision trees induced by a splitting algorithm (Rendell, 1983, 1985; Sikora Shaw, 1994) to expertsystem rules (Montana, 1990) to weights for a game s evaluation function (Rendell, 1990) to weights and orientations ....

Whitley, D., and J. D. Schaffer (eds.). 1992. Combinations of genetic algorithms and neural networks. Los Alamitos, Calif.: IEEE Computer Society Press.

....of learning parameters. Whitley et al. 110] discuss the use of genetic algorithms to optimize the connection weights in feedforward neural networks, and to discover novel architectures in the form of connectivity patterns for neural networks that learn using error propagation. Schaffer et al. [93] survey techniques that are based on genetic algorithms to evolve network parameters and learning rules. Fahlman and Lebiere [29] and Tenorio [101] present learning algorithms that construct the network topology during training, guided by a measure of performance. Techniques such as ....

....to avoid overfitting to the training data) and a testing data set. 100 9.4 Future Directions Future directions to this dissertation include: ffl Combining NNs generated by different learning techniques. A wide variety of learning algorithms and network architectures are currently available [28, 29, 45, 46, 86, 93]. Combining NNs generated using different learning techniques is of special value, since the collinearity among the errors of such networks may be small. This point leads to a very important research question: how to make the component NNs different or less collinear in order to increase the ....

J. D. Schaffer, D. Whitley, and L. J. Eshelman. Combinations of genetic algorithms and neural networks: A survey of the state of the art. In L. D. Whitley and J. D. Schaffer, editors, Proceedings of COGANN-92 International Workshop on Combinations of Genetic Algorithms and Neural Networks, pages 1--37. IEEE, 1992.

....combines learning with evolution and neural networks. The past decade has seen an explosion of the work on evolutionary neural networks, targeted mostly at automating the setting of parameters in feedforward neural networks (e.g. see the work of Mhlenbein and Kinderman (1989) the survey of Schaffer et al. (1992), the review of Yao (1993) the general framework proposed by Yao and Liu (1998) and the increasing number of papers and workshops addressing these issues) The work closest to ours, since it also considers RBF networks, is that of Neruda (1995) in which functionally equivalent canonical ....

Schaffer, J.D.; Whitley, D; Eshelman, L.J., 1992, "Combinations of Genetic Algorithms and Neural Networks: A Survey of the State of the Art", Proceedings of the International Workshop on Combinations of Genetic Algorithms and Neural Networks, 1-37.

....algorithms have been successfully combined in a second way which should not be confused with the model outlined above. Instead of evolving the parameters of the network, the parameters are now constant from one generation to the next. The genetic algorithm is used to evolve a network s weights (Schaffer 1992). The genetic algorithm is now the training algorithm. Although Schaffer et al. in their survey of past research note that genetic algorithms have been shown to be superior to backpropagation in both accuracy and speed of learning, many have had success combining this method with the previous ....

Schaffer, J.D., Whitley, D. and Eschelman, L.J. (1992). Combinations of genetic algorithms and neural networks: A survey of the state of the art. In Proceedings of COGANN-92, D. Whitley and J.D. Schaffer (Eds.), IEEE, Computer Society Press. U.S.A., June 6, 1992.

.... ICANN 97, Lausanne, Switzerland, October 1997, Proceedings, LNCS 1327, Springer Verlag, pp. 823 829. Although the combined application of neural network techniques and genetic algorithms turned out to be a very effective tool for solving an interesting class of problems (for a review see e.g. [10], 11] 1] we suggest, that the algorithm outlined in section 2 is better suited for the evolution of recurrent neural networks; especially in situations, where there is no good guess for an appropriate network architecture or where recurrent dynamic networks should be used for tasks like ....

Schaffer, J. D., Whitley, D., and Eshelman, L. J. (1992). Combination of genetic algorithms and neural networks: A survey of the state of the art. In: Proceedings International Workshop on combinations of genetic algorithms and neural networks (COGANN-92), Los Alamitos, CA, IEEE Computer Society Press.

....671, 672, 673] Elmasry, M.I. 134] El Sharkawi, M. A. 11, 442] Elsimary, Hamed, 493] Elsimary, H. 333] Enab, Y. M. 247] Enberg, Philippe Biela, 586] Engel, Paulo M. 156] English, T. M. 33] Ennaji, A. 479] Erives, H. 334] Ersoy, O. K. 71, 484] Eshelman, Larry J. [907, 908, 957] Esparcia Alcazar, Anna J. 494] Esparcia Alc azar, Anna I. 443] Estevez, Pablo A. 34, 108, 168] Eyvazova, Z. E. 222] Fadda, A. 169, 335] Fagg, Andrew H. 804] Falcon, J. F. 674] Fang, Jian, 512] Farag, A. A. 547] Fariselli, Piero, 224, 558] Fekadu, Adhanom A. 761] ....

....Sarabia, L. A. 548] Saratchandran, P. 170] 24 Genetic algorithms and neural networks Saravanan, N. 83, 282] Sarmiento, A. 32] Sase, M. 395] Sato, Y. 216, 422] Sato, Yuji, 84, 396] Saunders, Gregory M. 135] Saxena, Ashutosh, 110] Schafer, J. 217] Schaffer, J. David, [907, 908, 957] Scherer, A. 397] Schiffmann, Wolfram, 884, 909, 910, 911] Schizas, C. N. 837] Schizas, Christos N. 376] Schlageter, G. 397] Schlenzig, J. 686, 689] Schmeck, Hartmut, 19, 157, 190] Schmeck, Heinrich, 870] Schmidt, M. 398] Schneider, A. 283] Schneider, Gisbert, 38, ....

[Article contains additional citation context not shown here]

J. David Schaffer, Darrell Whitley, and Larry J. Eshelman. Combinations of genetic algorithms and neural networks: A survey of the state of the art. In Schaffer and Whitley [638], pages 1--37. y(CCA 58168/93) ga:Whitley92g.

....671, 672, 673] Elmasry, M.I. 134] El Sharkawi, M. A. 11, 442] Elsimary, Hamed, 493] Elsimary, H. 333] Enab, Y. M. 247] Enberg, Philippe Biela, 586] Engel, Paulo M. 156] English, T. M. 33] Ennaji, A. 479] Erives, H. 334] Ersoy, O. K. 71, 484] Eshelman, Larry J. [907, 908, 957] Esparcia Alcazar, Anna J. 494] Esparcia Alc azar, Anna I. 443] Estevez, Pablo A. 34, 108, 168] Eyvazova, Z. E. 222] Fadda, A. 169, 335] Fagg, Andrew H. 804] Falcon, J. F. 674] Fang, Jian, 512] Farag, A. A. 547] Fariselli, Piero, 224, 558] Fekadu, Adhanom A. 761] ....

....Sarabia, L. A. 548] Saratchandran, P. 170] 24 Genetic algorithms and neural networks Saravanan, N. 83, 282] Sarmiento, A. 32] Sase, M. 395] Sato, Y. 216, 422] Sato, Yuji, 84, 396] Saunders, Gregory M. 135] Saxena, Ashutosh, 110] Schafer, J. 217] Schaffer, J. David, [907, 908, 957] Scherer, A. 397] Schiffmann, Wolfram, 884, 909, 910, 911] Schizas, C. N. 837] Schizas, Christos N. 376] Schlageter, G. 397] Schlenzig, J. 686, 689] Schmeck, Hartmut, 19, 157, 190] Schmeck, Heinrich, 870] Schmidt, M. 398] Schneider, A. 283] Schneider, Gisbert, 38, ....

[Article contains additional citation context not shown here]

J. David Schaffer, Darrell Whitley, and Larry J. Eshelman. Combinations of genetic algorithms and neural networks: A survey of the state of the art. In Schaffer and Whitley [638], pages 1--37. y ga:Schaffer92a.

....Michael G. 236, 354] Dymek, A. 173] Ebeling, Werner, 176] Eberhart, R. C. 177] Eiben, Agoston E. 178] Eisenhammer, Thomas, 641] Elias, John G. 179, 180, 181] El Keib, A. A. 164, 165] Ellis, C. 182] English, Thomas M. 183] Esbensen, Henrik, 184] Eshelman, Larry J. [617, 618, 697] Fagg, Andrew H. 451] Falck, E. 79] Falco, I. De, 148] Falkenauer, Emanuel, 187] Fang, Hsiao Lan, 188] Feipeng, Li, 712] Feldberg, Rasmus, 577] Fenekohal, Rahim F. 264] Ferguson, J. J. 189] Fern andez, G. 675] Fieber, M. 418] Filelis, A. 548] Filipic, Bogdan, 190, ....

....F. 615] Sandqvist, Sam, 29] Sannomiya, Nobuo, 338, 12] Sano, Chiharu, 616] Santib a nez Koref, Ivan, 104, 105, 106, 107] Sappington, David E. 135] Sarma, Jayshree, 155] Sassus, F. 161] Sato, Taisuke, 279, 280, 281] Sato, T. 655] Savini, A. 268] Schaffer, J. David, [617, 618, 697] Schizas, C. N. 495] Schlenzig, J. 204] Schlierkamp Voosen, Dirk, 510, 511, 512] Schlosser, Steve G. 680] Schmid, L. J. 637] Schmitendorf, W. E. 218] Schoneburg, E. 310, 619] Schraudolph, Nicol N. 86, 87, 88, 273] Schulte, J. W. 18] Schultz, Alan C. 157] ....

[Article contains additional citation context not shown here]

J. David Schaffer, Darrell Whitley, and Larry J. Eshelman. Combinations of genetic algorithms and neural networks: A survey of the state of the art. In Schaffer and Whitley [115], pages 1--37. y(CCA 58168/93) ga:Whitley92g.

....Michael G. 236, 354] Dymek, A. 173] Ebeling, Werner, 176] Eberhart, R. C. 177] Eiben, Agoston E. 178] Eisenhammer, Thomas, 641] Elias, John G. 179, 180, 181] El Keib, A. A. 164, 165] Ellis, C. 182] English, Thomas M. 183] Esbensen, Henrik, 184] Eshelman, Larry J. [617, 618, 697] Fagg, Andrew H. 451] Falck, E. 79] Falco, I. De, 148] Falkenauer, Emanuel, 187] Fang, Hsiao Lan, 188] Feipeng, Li, 712] Feldberg, Rasmus, 577] Fenekohal, Rahim F. 264] Ferguson, J. J. 189] Fern andez, G. 675] Fieber, M. 418] Filelis, A. 548] Filipic, Bogdan, 190, ....

....F. 615] Sandqvist, Sam, 29] Sannomiya, Nobuo, 338, 12] Sano, Chiharu, 616] Santib a nez Koref, Ivan, 104, 105, 106, 107] Sappington, David E. 135] Sarma, Jayshree, 155] Sassus, F. 161] Sato, Taisuke, 279, 280, 281] Sato, T. 655] Savini, A. 268] Schaffer, J. David, [617, 618, 697] Schizas, C. N. 495] Schlenzig, J. 204] Schlierkamp Voosen, Dirk, 510, 511, 512] Schlosser, Steve G. 680] Schmid, L. J. 637] Schmitendorf, W. E. 218] Schoneburg, E. 310, 619] Schraudolph, Nicol N. 86, 87, 88, 273] Schulte, J. W. 18] Schultz, Alan C. 157] ....

[Article contains additional citation context not shown here]

J. David Schaffer, Darrell Whitley, and Larry J. Eshelman. Combinations of genetic algorithms and neural networks: A survey of the state of the art. In Schaffer and Whitley [115], pages 1--37. y ga:Schaffer92a.

....training schemes, the chromosome encodes the network and the fitness evaluation consists of constructing the network, testing the network, and obtaining a fitness assessment based on the performance of the network. Examples are provided in the overviews by Yao [80] Wei [75] and Schaffer et al. [67]. A useful early bibliography is given by Rudnick [63] The approaches taken can be broadly categorised as evolving fixed architecture neural networks, and evolving the network architecture itself. In both cases the activation function of each neuron is the same. These two approaches are ....

J. David Schaffer, Darrel Whitley, and Larry J. Eschelman. Combinations of genetic algorithms and neural networks: A survey of the state of the art. In COGANN'92: International Workshop on Combinations of Genetic Algorithms and Neural Networks, pages 1--37, Baltimore, MD, USA, 6 June 1992. IEEE Press.

....not imply that the solution is not a local optimum; rather, it eliminates the possibility of getting caught in local optima. Another appealing issue is the possibility of performing steps (a) and (b) at the same time, in a search over the joint space of structures and weights (for a review, see [22, 28]) Still another possibility is to embody a DBM into a GA, using the latter to search among the space of structures (a) and the DBM to optimize the weights (b) 16] this hybridization leads to extremely high computational costs. Finally, there is the use of EAs solely for the numerical ....

Shaffer, J.D., Whitley, D., Eshelman, L.J. Combination of Genetic Algorithms and Neural Networks: A Survey of the State of the Art. In Combination of Genetic Algorithms and Neural Networks. Shaffer, J.D., Whitley, D. (eds.), pp. 1-37, 1992.

....1 # ub = 1 # ub = 1 Algorithm 1 # 0 = 1.2 # 0 = 0.002 # 0 = 0.0034 Algorithm 2 aaa Algorithm 3 # 0 = 1.2 # 0 = 0.002 # 0 = 0.0034 # 0 = 15 # 0 = 1.5 # 0 = 1.5 # lb = 10 5 # lb = 10 5 # lb = 10 5 # ub = 1 # ub = 0.01 # ub = 1 a No heuristics required. (Schaffer, Whitley, Eshelman, 1992). Since the optimal value is highly dependent on the learning task, no general strategy has been developed to deal with this problem. Thus, the optimal value of m is experimental but depends on the learning rate chosen. In our experiments, we have tried nine different values for the momentum ....

Schaffer, J., Whitley, D., & Eshelman, L. (1992). Combinations of genetic algorithms and neural networks: A survey of the state of the art. In Proceedings of the International Workshop on Combinations of Genetic Algorithms and Neural Networks (pp. 1--37). Los Alamitos, CA: IEEE Computer Society Press.

No context found.

Schaffer, J.D., Whitley, D. & Eshelman, L.J. (1992). Combinations of genetic algorithms and neural networks: a survey of the state of the art. In D. Whitley and J.D. Schaffer (Eds.), Combinations of Genetic Algorithms and Neural Networks. IEEE Computer Society Press.

No context found.

J. Schaffer, D. Whitley, and L. Eshelman. Combinations of genetic algorithms and neural networks: A survey of the state of the art. In Proceedings of the International Workshop on Combinations of Genetic Algorithms and Neural Networks, pages 1--37, 1992.

No context found.

J. D. Schaffer, D. Whitley, and L. J. Eshelman. Combinations of genetic algorithms and neural networks: A survey of the state of the art. In Proceedings of COGANN-92 International Workshop on Combinations of Genetic Algorithms and Neural Networks, 1992.

No context found.

J. D. Schaffer, D. Whitley, and L. J. Eshelman, "Combinations of Genetic Algorithms and Neural Networks: A Survey of the State of the Art", International Workshop on Combinations of Genetic Algorithms and Neural Networks, Baltimore, Maryland, June 6, 1992, pp. 1-37.

No context found.

J. D. Shaffer, D. Whitley, and L. J. Eshelman, "Combinations of genetic algorithms and neural networks: A survey of the state of the art," in Proc. COGANN-92: Int. Workshop Combinations Genetic Algorithms Neural Networks, Los Alamitos, CA, 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