J. Merelo, M. Paton, A. Canas, A. Prieto, and F. Moran, "Optimization of a competitive learning neural network by Genetic Algorithms," Lecture Notes In Computer Science, vol. 686, pp. 185--192, 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....Horn et al. 25] and applied in the classification domain, to a synthetic two class problem. Other limited approaches to multi objective training do appear in the evolving topographies literature, but in the form of simultaneously choosing the network topology as well as Euclidean training e.g. [38] (these again however use the linear sum or penalty approach) The new framework proposed here is designed to use of those evolutionary computation (EC) methods which have previously been applied to uni objective NN design, genetic algorithms (GAs) evolution strategies (ES) and particle swarm ....

.... to use of those evolutionary computation (EC) methods which have previously been applied to uni objective NN design, genetic algorithms (GAs) evolution strategies (ES) and particle swarm optimisation (PSO) GAs have previously be used for feature selection [8, 53] and topography selection [2, 5, 29, 35, 36, 38, 52] and ESs have been used for weight optimisation [21, 42, 45, 55] and adaptive topography selection [15, 37, 57] The recent EC technique of PSO [27] has also proved popular as a uni objective NN optimiser [10, 12, 13, 26, 48] 2 Multi objective evolutionary neural network flamework The use of ....

J.J. Merelo, M. Paton, A. Canas, A. Prieto, and F. Moran. Optimization of a competitive learning neural network by Genetic Algorithms. Lecture Notes In Computer Science, 686:185-192, 1993.

....learning and evolution in biological organisms in order to understand their complex behavior. There are also studies in using evolutionary search to find an optimal learning parameter set for local search methods, such as the learning rate and momentum term in the backpropagation algorithm [27,38,53]. More ambitious works include the investigation of the evolution of local search methods [9,11,21] For example, the delta rule for feedforward neural networks has been successfully evolved in [9] 3 The Long term Dependency Problem Much of the previous work in combining gradient based ....

J. J. Merelo, M. Pat'on, A. Canas, A. Prieto, and F. Mor'an. Optimization of a competitive learning neural network by genetic algorithms. In Proceedings of the International Workshop on Artificial Neural Networks, pages 185--192, 1993.

....BP often has to run several times in practice in order to find good connection weights due to its sensitivity to initial conditions, the hybrid training algorithm will be quite competitive. Similar work on the evolution of initial weights has also been done on competitive learning neural networks [139] and Kohonen networks [140] It is interesting to consider finding good initial weights as locating a good region in the weight space. Defining that basin of attraction of a local minimum as being composed of all the points, sets of weights in this case, which can converge to the local minimum ....

....by Jacobs [273] because the simultaneous evolution of both algorithmic parameters and architectures facilitates exploration of interactions between the learning algorithm and architectures such that a near optimal combination of BP with an architecture can be found. Other researchers [32] [139], 213] 272] also used an evolutionary process to find parameters for BP but ANN s architecture was predefined. The parameters evolved in this case tend to be optimized toward the architecture rather than being generally applicable to learning. There are a number of BP algorithms with an ....

J. J. Merelo, M. Pat on, A. Ca nas, A. Prieto, and F. Mor an, "Optimization of a competitive learning neural network by genetic algorithms," in Proc. Int. Workshop Artificial Neural Networks (IWANN'93), Lecture Notes in Computer Science, vol. 686. Berlin, Germany: Springer-Verlag, 1993, pp. 185--192.

....multiple error measures. A method of error smoothing is also introduced as an attempt to solve the current stopping problem associated with ES (and GA) trained NNs. The use of Evolutionary and Genetic approaches to Neural Network training has received increasing attention in recent years [4, 5, 6, 7, 9, 12, 14]. Indeed the ability of these approaches to facilitate NN training beyond the Euclidean objective was highlighted by Porto et al. [9] but apparently taken no further. Other limited approaches to multi objective training do appear in the literature, but in the form of simultaneously choosing the ....

....NN training beyond the Euclidean objective was highlighted by Porto et al. [9] but apparently taken no further. Other limited approaches to multi objective training do appear in the literature, but in the form of simultaneously choosing the network topology as well as Euclidean training e.g. [7] 5 Diagram 1, generic population selection. The ES proposed for use in this multi objective learning model is that used in [4, 9, 12] in the latter two it is referred to as Evolutionary Programming) and is shown in Eq. 3. Here the weight space of a network is perturbed by some values drawn from ....

Merelo, J.J., Paton, M., Canas, A, Prieto, A and Moran, F., "Optimization of a competitive learning neural network by Genetic Algorithms", Lecture Notes In Computer Scince, Vol. 686, pp185-192, 1993.

....Richard, 577] Bushnell, M. J. 776] Bustillo, Eduardo, 491] Butler, Darren, 516] Buydens, Lutgarde M. C. 802] Byrne, J. A. 857] Calabretta, R. 321] Caloba, L. P. 317] Caloba, L. P. 480] Campadelli, P. 319] Campanini, Renato, 224] Campos, M. F. M. 195] Canas, A. [865, 866] Caponetto, R. 639, 640] Capozza, M. 592] Carazo, J. M. 513] Card, H. C. 475] Carpentieri, M. 319] Carrier, Jean Yves, 322] Carse, Brian, 241, 436, 471, 474] Caruana, Richard A. 907] Casadio, Rita, 224, 558] Caskey, Kevin Richard, 641] Caudell, Thomas P. 642, 643] ....

....John, 614, 615, 616] McInerney, Michael, 805] McInerney, M. 62] Mecklenburg, Klaus, 910] Meeden, Lisa A. 455] Meisel, J. 417] Melsheimer, S. S. 55] Menczer, Filippo, 834, 835, 836] Mendoca, P. R. S. 480] Mendonca, P. R. S. 317] Meng, Qing chun, 277] Merelo, J. J. [153, 198, 293, 513, 865, 866] Meservy, R. D. 346] Meyer, Claudia M. 867, 868] Meyer, Jean Arcady, 772] Michel, Olivier, 200] Middleton, L. T. 837] Miglino, Orazio, 63, 296, 862] Mihaila, D. 298] Miikkulainen, Risto, 118, 838, 839] Mikami, Sadayoshi, 553, 574] Miller, Geoffrey F. 37, 711, 712, 713] ....

[Article contains additional citation context not shown here]

J. J. Merelo, A. Paton, A. Canas, A. Prieto, and F. Moran. Optimization of a competitive learning neural network by genetic algorithms. In Proceedings of the International Workshop on Artificial Neural Networks (IWANN'93), pages 185--192, Sitges (Spain), 9.-11. June 1993. Springer-Verlag, Berlin. y(CCA 17308/93) ga:Paton93b.

....Brandao, M. A. 41] Braunstingl, R. 163] Bretas, Newton G. 56] Buckles, Bill P. 103, 104] Bustillo, Eduardo, 212] Cadenas, Jose Manuel, 143, 146, 166] Cain, G. 192] Caloba, L. P. 39, 55] Camacho, E. F. 158, 230] Campodonico, N. M. 37] Camponogara, Eduardo, 57] Canas, A. [250] Candido, M. A. B. 58] Carazo, J. M. 223] Carfalhode, A. 60] Carreno, D. 210] Carvalho, Andr e C. P. L. F. de, 56] Carvalho, Luis, 116] Carvalho, P. M. S. 120, 126, 127] Casao, Jorge Gasos, 244] Castellanos, J. 236] Castillo, Luis, 237] Castro, J. L. 248] Castro, Jesus Silva, ....

....Javier, 138, 145, 251, 252] Marques, R. M. Lopes, 132] Martin, C. A. 21, 27] Martin, F. J. Marin, 173] Martin Bautista, M. J. 228] Martins, Weber, 29] McInnes, F. R. 20] Mendes, R. 112] Mendoca, P. R. S. 55] Mendonca, P. R. S. 39] Mercedes, de Las, 216] Merelo, J. J. [155, 182, 193, 222, 223, 250] Michaelian, K. 91, 101] Miliani, L. 265] Miranda, V. 113, 115, 116, 117, 123, 125, 64] Monasterio, F. 154, 167] Monasterio Huelin, F. 234] Montenegro, Anselmo A, 31] Montero, G. 135, 142] Montilla, Guillermo, 265, 266] Morales, D. 162] Moran, F. 223, 238, 250] Moreira, D. de ....

[Article contains additional citation context not shown here]

J. J. Merelo, A. Paton, A. Canas, A. Prieto, and F. Moran. Optimization of a competitive learning neural network by genetic algorithms. In Proceedings of the International Workshop on Artificial Neural Networks (IWANN'93), pages 185--192, Sitges (Spain), 9.-11. June 1993. Springer-Verlag, Berlin. yCCA 17308/93 ga:Paton93b.

....R. 102, 104, 105, 106] Bull, Lawrence, 290] Bullock, G. N. 795] Burghof, Axel, 148] Burks, Christian, 306, 307] Bushnell, M. J. 504] Buydens, Lutgarde M. C. 641, 642, 643, 644] Cai, H. 517] Cai, J. 1008] Cain, G. D. 152, 153, 154, 155] Campbell, J. A. 498] Canas, A. [802, 803] Caponetto, R. 156, 157, 158] Carbonaro, Antonella, 165] Carlson, Susan Elizabeth, 159, 934, 935] Carpenter, Tamra, 562] Carrick, C. 160] Carter, Bob, 537] Cartwright, Hugh M. 161, 162, 163] Casadei, Giorgio, 164, 165] Casao, Jorge Gasos, 146] Caskey, Kevin Richard, 166] ....

....D. B. 524] McGraw, G. 637] McGregor, Douglas R. 704, 705, 706, 707, 708, 709] McInerney, Michael, 657] Medsker, C. 710] Mellish, C. 270] Mendes, E. M. 287, 305] Meng, Qing chun, 426, 711] Menozzi, J. J. 839] Menozzi, John J. 697] Menth, Stefan, 67] Merelo, J. J. [802, 803] Merkle, Laurence D. 712] Merz, Jr. K. M. 619] Meyer, Claudia M. 806] Meyer, Robert R. 843, 844] Michalewicz, Zbigniew, 713, 714, 715, 716] Michielsen, E. 717] Michielssen, E. 841, 842] Miglino, Orazio, 791] Miikkulainen, Risto, 718] Mikami, Sadayoshi, 543, 545] Miles, ....

[Article contains additional citation context not shown here]

J. J. Merelo, A. Paton, A. Canas, A. Prieto, and F. Moran. Optimization of a competitive learning neural network by genetic algorithms. In Proceedings of the International Workshop on Artificial Neural Networks (IWANN'93), pages 185--192, Sitges (Spain), 9.-11. June 1993. Springer-Verlag, Berlin. y(CCA 17308/93) ga:Paton93b.

....BP often has to run several times in practice in order to find good connection weights due to its sensitivity to initial conditions, the hybrid training algorithm will be quite competitive. Similar work on the evolution of initial weights has also been done on competitive learning neural networks [139] and Kohonen networks [140] It is interesting to consider finding good initial weights as locating a good region in the weight space. Defining that basin of attraction of a local minimum as being composed of all the points, sets of weights in this case, which can converge to the local minimum ....

....to parents to generate offspring which form the new generation. Figure 12: A typical cycle of the evolution of learning rules. exploration of interactions between the learning algorithm and architectures such that a nearoptimal combination of BP with an architecture can be found. Other researchers [32, 139, 213, 272] also used an evolutionary process to find parameters for BP but ANN s architecture was predefined. The parameters evolved in this case tend to be optimized towards the architecture rather than being generally applicable to learning. There are a number of BP algorithms with an adaptive learning ....

J. J. Merelo, M. Pat'on, A. Ca~nas, A. Prieto, and F. Mor'an, "Optimization of a competitive learning neural network by genetic algorithms," in Proc. of Int'l Workshop on Artificial Neural Networks (IWANN'93), pp. 185--192, Springer-Verlag, 1993. Lecture Notes in Computer Science, Vol. 686.

....an appropriate number of nodes in each layer, the intermediate level to find a suitable connectivity and the lowest level to set the weights of the network. Each level uses the next lower level for evaluation which of course makes the whole procedure extremely time consuming. Merelo et al. [50] devised a two layer network for classification problems where the first (hidden) layer is trained following a competitive learning algorithm and the second layer is trained by perceptron learning. The genetic algorithm is used to find learning parameters, the number of units in the first layer ....

J. J. Merelo, M. Paton, A. Canas, A. Prieto, and F. Moran. Optimization of a competitive learning neural network by genetic algorithms. In J. Mira, J. Cabestany, and A. Prieto, editors, Proceedings of the International Work shop on Artifical Neural Networks, pages 185--192. Springer-Verlag, June 1993.

....to find good connection weights because of BP s sensitivity to initial conditions, the hybrid training algorithm is quite competitive in comparison with gradient based training algorithms. Similar work on the evolution of initial weights has also been done on competitive learning neural networks [73]. It is interesting to consider finding good initial weights as locating a good region in the space. Defining that the basin of attraction of a local minimum is composed of all the points, sets of weights in this case, which can converge to the local minimum through a local search algorithm, then ....

....of interactions between the learning algorithm and architectures so that a near optimal combination of a BP algorithm with an architecture can be evolved. The cost of this benefit, as mentioned in Section 3.4, is a larger search space and thus longer computation time. Some other researchers [35, 73] also used an evolutionary process to find parameters for the BP algorithm, but EANN s architecture was pre defined. The parameters evolved in this case tend to be optimised towards the architecture, rather than general applicable ones. There are also a number of BP algorithms with an adaptive ....

J. J. Merelo, M. Pat'on, A. Ca~nas, A. Prieto, and F. Mor'an. Optimization of a competitive learning neural network by genetic algorithms. In Proc. of Int'l Workshop on Artificial Neural Networks (IWANN'93), pages 185--192. Springer-Verlag, 1993. Lecture Notes in Computer Science, Vol. 686.

.... descent algorithm, designed to minimize, step by step, the di erence between the actual output vector of the network and the desired output vector, such as BP in its di erent versions (like, for instance QuickProp, QP, by [1] and RPROP by Riedmiller and Braun [2] Some evolutionary approaches [3 7] are also used. However, this method, successful as it is in many elds, does encounter certain diculties in practice: 1) the convergence tends to be extremely slow; 2) convergence to the global optimum is not guaranteed; 3) learning constants and other parameters must be arrived at ....

.... methods of indirect coding have been proposed, such as those of Kitano [30] Harp et al. 31] Dodd et al. 32] and Gruau [33] Search for the optimal learning parameters, including weights, having preestablished the number of neurons and the connectivity between them, as did Merelo et al. in [3], for multilayer competitive learning neural nets, where a method that codi es the weights and learning parameters onto chromosomes is presented. Another example is the method presented by Petridis et al. in [34] where both weights and learning parameters are represented as bit strings to be ....

J.J. Merelo; M. Paton; A. Ca~nas; A. Prieto and F. Moran. Optimization of a competitive learning neural network by genetic algorithms. IWANN93. Lectures Notes in Computer Science, vol. 686, 185-192, 1993.

....and the desired output vector, such as BP in its di erent versions (like, for instance QuickProp by Fahlman et Preprint submitted to Elsevier Preprint 25 April 2000 al. 1] and RPROP by Riedmiller and Braun [2,3] is widely used as training mechanism, as well as other evolutionary approaches [4 8]. However, even as these BP methods are successfully used in many elds, especially for pattern recognition, due to its learning ability, it does encounter certain diculties in practice: 1) convergence tends to be extremely slow and very dependent on the initial weights; 2) convergence to the ....

.... some methods of indirect coding have been proposed, as did Kitano [34] Harp et al. 35] Dodd et al. 36] and Gruau [37] Search for the optimal learning parameters, including weights, having preestablished the number of neurons and the connectivity between them, as did Merelo et al. in [4], for multilayer competitive learning neural nets, where a method that codi es the weights and learning parameters onto chromosomes is presented. Another example is the method presented by Petridis et al. in [38] where both weights and learning parameters are represented as bit strings to be ....

J.J. Merelo; M. Paton; A. Ca~nas; A. Prieto and F. Moran. Optimization of a competitive learning neural network by genetic algorithms. Lectures Notes in Computer Science, Springer Verlag, vol. 686, 185-192, 1993.

....(structured in layers and correctly connected) their initial parameters established and then trained. BP in its different versions is widely used as training mechanism; examples include QuickProp (QP) 1] and RPROP (Riedmiller and Braun [2] as well as other evolutionary based approaches [3, 4, 5, 6]. Whichever method is chosen, the training mechanism is an iterative gradient descent algorithm designed to step by step minimize the difference between the actual output vector of the network and the desired output vector. Although this method is used successfully in many fields, especially for ....

....approach to face the optimization of the design of a neural network architecture and the choice of the best learning method. Other approaches search for the optimal learning parameters, having pre established the number of neurons and the connectivity between them, as did Merelo et al. in [3], Petridis et al. in [23] or Castillo et al. in [24] where an approach based on Simulated Annealing and BP is presented. Both decremental and incremental algorithms are gradient descent optimization methods, so they suffer the problem that they may reach the closest local minimum to the search ....

J.J. Merelo; M. Paton; A. Canas; A. Prieto and F. Moran. Optimization of a competitive learning neural network by genetic algorithms. IWANN93. Lectures Notes in Computer Science, vol. 686, 185-192, 1993.

....approach to face the optimization of the design of a neural network architecture and the choice of the best learning method. ffl Other approaches search for the optimal learning parameters, having pre established the number of neurons and the connectivity between them, as did Merelo et al. in [22], Petridis et al. in [23] or Castillo et al. in [24] where an approach based on Simulated Annealing and BP is presented. ffl A method to search for the optimal set of weights, the optimal topology and learning parameter, using a GA and BP was proposed by Castillo et al. in [25] however the ....

J.J. Merelo; M. Pat'on; A. Canas; A. Prieto and F. Mor'an. Optimization of a competitive learning neural network by genetic algorithms. IWANN93. Lectures Notes in Computer Science, vol. 686, 185-192, 1993.

No context found.

J. Merelo, M. Paton, A. Canas, A. Prieto, and F. Moran, "Optimization of a competitive learning neural network by Genetic Algorithms," Lecture Notes In Computer Science, vol. 686, pp. 185--192, 1993.

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