F. Menczer and D. Parisi, "Evidence of hyperplanes in the genetic learning of neural networks," Biological Cybernetics, vol. 66, pp. 283--289, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....generation by forming a linear regression model relating the fitness of a crossover produced child to the average fitness of the two parents, and then calculating correlation coe#cients that measure the degree to which the linear regression model explains the observed data. Menczer and Parisi [16] compared the performance of GAs with and without crossover in the optimisation of weights in a neural net, and found that crossover did make a significant contribution to the GA s performance. They conducted experiments measuring the correlation coe#cients for both the mutation and crossover ....

F. Menczer and D. Parisi, "Evidence of hyperplanes in the genetic learning of neural networks," Biological Cybernetics, vol. 66, pp. 283--289, 1992.

....a changing environment. In the literature, research into EANN has been taking one of three approaches; evolving the weights of the network, evolving the architecture, or evolving both simultaneously. The EANN approach uses either binary representation to evolve the weight matrix [12, 13] or real [7, 8, 9, 18, 19, 23]. There is not an obvious advantage of binary encoding in EANN over the real. However, with real encoding, there are more advantages including compact and natural representation. The key problem (other than being trapped in a local minimum) with BP and other traditional training algorithms is the ....

F. Menczer and D. Parisi. Evidence of hyperplanes in the genetic learning of neural networks. Biological Cybernetics, 66:283--289, 1992.

....a changing environment. In the literature, research into EANN has been taking one of three approaches; evolving the weights of the network, evolving the architecture, or evolving both simultaneously. The EANN approach uses either binary representation to evolve the weight matrix [10, 11] or real [6, 7, 16, 19]. There is not an obvious advantage of binary encoding in EANN over the real. However, with real encoding, there are more advantages including compact and natural representation. The key problem (other than being trapped in a local minimum) with BP and other traditional training algorithms is the ....

F. Menczer and D. Parisi. Evidence of hyperplanes in the genetic learning of neural networks. Biological Cybernetics, 66:283--289, 1992.

....Generally, a population of candidate solutions, ranked by their performance, are maintained and updated iteratively by evolutionary search. Each candidate solution represents one neural network in which the weights can be encoded as a string of binary [14,81,82] or floatingpoint numbers [24,52,66,71]. The performance of each solution is determined by the network error function which is to be optimized by evolutionary search. Unlike local search, evolutionary search maintains a population of potential solutions rather than a single solution. Therefore, the risk of getting stuck in local optima ....

F. Menczer and D. Parisi. Evidence of hyperplanes in the genetic learning of neural networks. Biological Cybernetics, 66:283--289, 1992.

....makes crossover operator very inefficient and ineffective in producing good offspring. B. Real Number Representation There have been some debates on the cardinality of the genotype alphabet. Some have argued that the minimal cardinality, i.e. the binary representation, might not be the best [48], 114] Formal analysis of nonstandard representations and operators based on the concept of equivalent classes [115] 116] has given representations other than ary strings a more solid theoretical foundation. Real numbers have been proposed to represent connection weights directly, i.e. one ....

....and operators based on the concept of equivalent classes [115] 116] has given representations other than ary strings a more solid theoretical foundation. Real numbers have been proposed to represent connection weights directly, i.e. one real number per connection weight [27] 29] 30] [48], 63] 65] 74] 95] 96] 102] 110] 111] 117] 118] For example, a realnumber representation of the ANN given by Fig. 3(a) could be (4.0,10.0,2.0,0.0,7.0,3.0) As connection weights are represented by real numbers, each individual in an evolving population will be a real vector. ....

[Article contains additional citation context not shown here]

F. Menczer and D. Parisi, "Evidence of hyperplanes in the genetic learning of neural networks," Biological Cybern., vol. 66, pp. 283--289, 1992.

....Int. Artif. Intell. Neural Netw. Complex Probl. Solving Technol. Netherlands) 467] Applied Mathematics and Computation, 590] Artif. Intell. Eng. UK) 424, 512] Artificial Life, 63, 301, 305, 556] Autom. Electr. Power Syst. China) 488] Bioinformatics, 588] Biological Cybernetics, [681, 835] Biophysical Journal, 126, 173] Bull. Fac. Eng. Univ. Tokushima (Japan) 336] Bull. Sci. Assoc. Ing. Electr. Inst. Electrotech. Montefiore, 633] Bulletin of the Polish Academy of Sciences Chemistry, 100] Cancer Letters, 313] Chemometrics and Intelligent Laboratory Systems, 149, 495, ....

....John R. 61, 117, 822, 823, 824, 825, 826, 827, 828] McGregor, Douglas R. 829, 830, 831, 832, 833] McInerney, 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, ....

[Article contains additional citation context not shown here]

Filippo Menczer and Domenico Parisi. Evidence of hyperplanes in the genetic learning of neural networks. Biological Cybernetics, 66(3):283--289, 1992. ga:Menczer92a.

.... Acta, 212, 384, 448, 449] Annals of Mathematics and Artificial Intelligence, 153, 265, 682, 683, 690, 691] APL Quote Quad, 392] Applied Artificial Intelligence, 258, 587] Archiv fur Elektrotechnik, 79] Atoms, Molecules and Clusters, 418] BioEngineering, 223] Biological Cybernetics, [57, 488] Biopolymers, 447] CC AI, 199] Clinical Chemistry, 308] Complex Systems, 139, 222, 245, 247, 248, 326, 509, 516, 532, 540, 595, 597] Composites Engineering, 116] Computer Physics Communications, 571] Computers Industrial Engineering, 524, 546] Computers Mathematics with ....

....[670] Maza, Michael de la, 473] Mazer, Emmanuel, 656, 658, 660] Mazumder, Pinaki, 474] McCaskill, J. S. 475] McCormick, Vance E. 325] McDonnell, John R. 476, 477, 478, 479, 480, 481] McGregor, Douglas R. 482, 483, 484, 485, 486, 487] McInerney, John, 86, 87] Menczer, Filippo, [488, 489] Meredith, D. L. 374] Merkle, Laurence D. 490] Messa, Kenneth, 492] Messa, K. 491] Meyer, Fred, 185] Meygret, A. 589] Meza, J. C. 360] Michalewicz, Zbigniew, 353, 493] Michielssen, E. 569] Middleton, L. T. 495] Miikkulainen, Risto, 496] Mikami, Sadayoshi, 368] ....

[Article contains additional citation context not shown here]

Filippo Menczer and Domenico Parisi. Evidence of hyperplanes in the genetic learning of neural networks. Biological Cybernetics, 66(3):283--289, 1992. ga:Menczer92a.

....or even continuous since EAs do not depend on gradient information. Because EAs can treat large, complex, nondifferentiable and multimodal spaces, which are the typical case in the real world, considerable research and application has been conducted on the evolution of connection weights [24, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112]. The evolutionary approach to weight training in ANNs consists of two major phases. The first phase is to decide the representation of connection weights, i.e. whether in the form of binary strings or not. The second one is the evolutionary process simulated by an EA, in which search operators ....

....makes crossover operator very inefficient and ineffective in producing good offspring. 2.2 Real Number Representation There have been some debates on the cardinality of the genotype alphabet. Some have argued that the minimal cardinality, i.e. the binary representation, might not be the best [48, 114]. Formal analysis of nonstandard representations and operators based on the concept of equivalent classes [115, 116] has given representations other than k ary strings a more solid theoretical foundation. Real numbers have been proposed to represent connection weights directly, i.e. one real ....

[Article contains additional citation context not shown here]

F. Menczer and D. Parisi, "Evidence of hyperplanes in the genetic learning of neural networks, " Biological Cybernetics, vol. 66, pp. 283--289, 1992.

....tests using NK landscapes. As we would expect, their results show a decreasing correlation coefficient for the crossover operator and increasing difficulty for the GA as the degree of epistatic interaction (maximum order of non zero partition coefficients) is increased. 9 Menczer and Parisi [35] compared the performance of GAs with and without crossover in the optimisation of weights in a neural net. They found that crossover did make a significant contribution to the GA s performance. They quote work arguing that for any local optimisation technique to make progress the fitness ....

Menczer, F., and Parisi, D., "Evidence of Hyperplanes in the Genetic Learning of Neural Networks," Biological Cybernetics, 66, 283-289 (1992)

....since GAs do not depend on gradient information in search. Because GAs are good at dealing with large, complex, nondifferentiable and multimodal spaces which are the typical space defined by an error function or fitness function, a lot of work has been done on the evolution of connection weights [29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64]. The evolutionary approach to weight training in EANNs consists of two major stages. The first stage is to decide the genotype representation of connection weights, i.e. whether in the form of binary strings or not. The second one is the evolution itself simulated by a GA or other evolutionary ....

....large EANNs will become extremely long and the evolution very inefficient. 2.2 Real Number Representation There have been some debates on the cardinality of the genotype alphabet. Some researchers argued that the minimal cardinality, i.e. the binary representation, might not be the best [52, 66]. Formal analysis of nonstandard representations and operators based on the concept of equivalent classes [67] has given representations other than k ary strings a more solid theoretical foundation. Real numbers were proposed to represent connection weights directly, i.e. one real number per ....

[Article contains additional citation context not shown here]

F. Menczer and D. Parisi. Evidence of hyperplanes in the genetic learning of neural networks. Biological Cybernetics, 66:283--289, 1992.

....its fitness. 4. Apply genetic operators, such as crossover and mutation, to each child individual generated above and obtain the next generation. Figure 1: A typical cycle of the evolution of connection weights. fitness function, a lot of work has been done on the evolution of connection weights [27, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62]. The evolution of connection weights provides an alternative approach to training EANNs. Such an evolutionary approach consists of two major stages. The first stage is to decide the genotype representation of connection weights, i.e. whether in the form of binary strings or not. The second one ....

....large EANNs will become extremely long and the evolution very inefficient. 2.2 Real Number Representation There have been some debates on the cardinality of the genotype alphabet. Some researchers argued that the minimal cardinality, i.e. the binary representation, might not be the best [51, 64]. Real numbers themselves were also proposed to represent connection weights, i.e. one real number per connection weight [30, 32, 33, 51] An EANN is represented by a set of real numbers, instead of binary strings. A recombination operator can exchange real numbers between two sets, but cannot ....

[Article contains additional citation context not shown here]

F. Menczer and D. Parisi. Evidence of hyperplanes in the genetic learning of neural networks. Biological Cybernetics, 66:283--289, 1992.

....on new results in feature selection for classification. 1. 1 Introduction The synthesis of neural architectures has been among the earliest applications of evolutionary computation [60, 1, 50, 13] Evolutionary algorithms have been used to adjust the weights of neural networks without supervision [51, 46], to design neural architectures [49, 28, 20, 48] and to find learning rules [5] Evolutionary algorithms, however, typically lead to uniform populations. This was appropriate in the above applications, since some optimal solution was assumed to exist. However, homogeneous solutions neural ....

....with another member of the population before being evaluated or before being inserted into the population, in case the parent is selected for reproduction. There has been ample discussion in the literature about the feasibility of recombination in the evolution of neural networks (see, e.g. [47, 46]) but such discussion is outside the scope of this chapter. Mutation is the only genetic operator used in the domains discussed in this chapter. The mutation operator provides evolving neural nets with a local search step, and the details of its dynamics are discussed for each task specific ....

F Menczer and D Parisi. Evidence of hyperplanes in the genetic learning of neural networks. Biological Cybernetics, 66:283--289, 1992.

....a new field called computational neuro ethology. Parisi et al. 18] have used feed forward neural networks called econets to simulate the sensory motor systems of very simple organisms which move in environments. Other authors have used similar approaches analyzing more complex sensory systems [2, 15, 16]. Our present work falls within the framework of these models. We imagine the TSP problem in terms of an organism that is inside an environment and must reach all target points scattered in it. At any given time, the organism only accesses (local) information about its surroundings by means of a ....

Menczer F and Parisi D, Evidence of Hyperplanes in the Genetic Learning of Neural Networks. Biol Cybern 66 (1992) 283-289.

No context found.

F. Menczer and D. Parisi. Evidence of Hyperplanes in the Genetic Learning of Neural Networks. Biological Cybernetics, 66(3):283--289, 1992.

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