X. Yao and Y. Liu, "Making Use of Population Information in Evolutionary Neural Networks," IEEE Transactions on Systems, Man and Cybernetics - Part B: Cybernetics, vol. 28, no. 3, pp. 417--425, 1998.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....and natural representation. The key problem (other than being trapped in a local minimum) with BP and other traditional training algorithms is the choice of a correct architecture (number of hidden nodes and connections) This problem has been tackled by the evolutionary approach in many studies [4, 14, 17, 21, 24, 39, 40, 41]. In some of these studies, weights and architectures are evolved simultaneously. The major disadvantage to the EANN approach is it is computationally expensive, as the evolutionary approach is normally slow. To overcome the slow convergence of the evolutionary approach to ANN, hybrid techniques ....

X. Yao and Y. Liu. Making use of population information in evolutionary artificial neural networks. IEEE Trans. on Systems, Man, and Cybernetics, Part B: Cybernetics, 28(3):417-- 425, 1998.

....and natural representation. The key problem (other than being trapped in a local minimum) with BP and other traditional training algorithms is the choice of a correct architecture (number of hidden nodes and connections) This problem has been tackled by the evolutionary approach in many studies [12, 15, 20, 30, 31]. In some of these studies, weights and architectures were evolved simultaneously. The major disadvantage to the EANN approach is it is computationally expensive, as the evolutionary approach is normally slow. To overcome the slow convergence of the evolutionary approach to ANN, hybrid techniques ....

X. Yao and Y. Liu. Making use of population information in evolutionary artificial neural networks. IEEE Trans. on Systems, Man, and Cybernetics, Part B: Cybernetics, 28(3):417--425, 1998.

.... 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 evolutionary approaches to NN training (with a single error function) has received increasing attention in recent ....

X. Yao and Y. Liu. Making Use of Population Information in Evolutionary Neural Networks. IEEE Transac- tions on Systems, Man and Cybernetics - Part B: Cybernetics, 28(3):417-425, 1998.

....fitness in P (i) Apply reproduction operators to the parents and produce offspring, the next generation, P(i 1) is obtained from the offspring and possibly parents. 18 work of manually finding an optimal network [2] 6] 7] 14] 18] 19] 20] 21] 37] 50] 60] 69] 70] 77] 81] 83] [84] [85] The advantage of the automatic design over the manual design becomes clearer as the complexity of ANN increases. EANNs provide a general framework for investigating various aspects of simulated evolution and learning [10] 14] 15] 50] 52] 3.2.1 General Framework for EANNs In EANN s ....

Yao X and Liu Y (1998), Making use of population information in evolutionary artificial neural networks, IEEE Transactions on Systems, Man and Cybernetics, Part B: Cybernetics, 28(3): PP.417-425.

....a very active area and there are a large number of research groups working on it. Besides the achievements cited in the brief review presented in Section 1, some significant developments in this area can be found in [Sharkey, 1999] Moreover, there are some works very related to this paper. Yao and Liu [1998] employed genetic algorithm to evolve a population of neural networks. Instead of choosing the best neural network in the last generalization as the final result, they regarded the entire population as a neural network ensemble and combining all the individuals in the last generalization in order ....

X. Yao and Y. Liu. Making use of population information in evolutionary artificial neural networks. IEEE Trans. Systems, Man and Cybernetics - Part B: Cybernetics, 28(3): 417-425, 1998.

....10, 7] Usually, neural networks are combined after training and are hence already quite perfect in solving a classification or approximation problem on their own. The ensemble members are not trained in combination and the composition of the ensemble does not undergo an optimization process. In [18] neural networks are evolved and a subset of the final population is combined afterwards. Di#erent combination methods including averaging and majority voting are compared while a genetic algorithm is used to search for a near optimal ensemble composition. For genetic programming Zhang et al. ....

X. Yao and Y. Liu, Making use of population information in evolutionary artificial neural networks. IEEE Transactions on Systems, Man and Cybernetics, 28B(3):417-- 425, 1998.

....parameters. For the evolutionary search of architectures, it will be interesting to model as co evolving sub networks [17] instead of evolving the whole network. Further, it will be worthwhile to explore the whole population information of the final generation for deciding the best solution [19]. We used a fixed chromosome structure (direct encoding technique) to represent the connection weights, architecture, learning algorithms and its parameters. As size of the network increases, the chromosome size grows. Moreover, implementation of crossover is often difficult due to production ....

Yao X and Liu Y, Making use of population information in evolutionary artificial neural networks, IEEE Transactions on Systems, Man and Cybernetics, Part B: Cybernetics, 28(3): pp.417425, 1998.

....hidden units. Maclin and Shavlik [25] initialize individual networks at di#erent points of the weight space. Krogh and Vedelsby [23] employ cross validation to create individual networks. Opitz and Shavlik [28] exploit genetic algorithm to train diverse knowledge based neural networks. Yao and Liu [46] ensemble all the individuals in an evolved population of neural networks. Zhou et al. 49] employs genetic algorithm to select a subset of a population of neural networks. As for combining the predictions of individual neural networks, the approaches used for classification and regression are ....

X. Yao and Y. Liu, Making use of population information in evolutionary artificial neural networks, IEEE Transactions on Systems, Man and Cybernetics - Part B: Cybernetics 28 (1998): 417-425.

.... a very weak allocator, or simply delete the allocator, we have the neural network ensemble (NNE) Usually, each module is a neural network designed using ffl different feature set, ffl different training set, ffl different initial condition, ffl different training algorithms, and so on [11] [16]. After designing each module separately, they are integrated together using a coordinator. What a coordinator does is very simple: just averaging the results of all modules usually provides a good final result. 5. One Class One Network If the modules are strong enough to know what to do, both ....

....problem. The data used here is the optdigits data set, which is taken from machine learning repository of the University of California at Irvine. The number of training examples is 3823, and the number of test examples is 1797. The number of features is 64 with each feature being an integer in [0,16]. The number of class is 10 (10 digits) Detailed information can be found in the file optdigits.names, which is included in the data set. The main experiment parameters are as follows. First, for each E module, the number of output neurons is 2, the number of hidden neurons is 4, and the number ....

X. Yao and Y. Liu, "Making use of population information in evolutionary artificial neural networks, " IEEE Trans. on Systems, Man, and Cybernetics - Part B, Vol. 28, No. 3, pp. 417-425, 1998.

....Experiments with the UCI benchmark datasets show that the proposed methods can produce more speciated ANNs and, thus, improve the performance by combining representative individuals with combination methods. 1 Introduction Combining multiple evolved ANNs has been actively researched recently [11, 14, 17]. Generally, multiple ANNs in the last generation are combined to construct an ensemble that has better generalization performance provided that the components, i.e. ANNs, complement each other in generalization. In terms of ANN, it means that an ANN produces does not have no or little common ....

X. Yao and Y. Liu, "Making use of population information in evolutionary artificial neural networks," IEEE Transactions on Systems, Man and Cybernetics, Part B: Cybernetics, vol. 28, no. 3, pp. 417-425, June 1998.

....that uses genetic algorithms to search for a correct and diverse population of neural networks to be used in the ensemble. It has an objective function that measures both the accuracy of the network and the disagreement of that network with respect to the other members of the set. Yao and Liu [3] experimented with a variety of combination methods to integrate multilayer perceptrons that were evolved by evolutionary programming and a modi ed back propagation algorithm. They also try to nd a committee of variable size using a genetic algorithm. However, only the individuals in the nal ....

Yao, X., Liu, Y.: Making Use of Population Information in Evolutionary Articial Neural Networks. IEEE Transactions on Systems, Man, and Cybernetics, 28B(2) (1998) 417-425. Byoung-Tak Zhang and Je-Gun Joung

.... evolutionary search of architectures, it will be also interesting to model as co evolving sub networks instead of evolving the whole network as we evolved in ALEC [9] Also it will be worthwhile to explore the whole population information of the final generation for deciding the best solution [8]. ....

Yao X and Liu Y, Making Use of Population Information in Evolutionary Artificial Neural Networks , IEEE Transactions on Systems, Man and Cybernetics, Part B: Cybernetics , 28(3): 417-425, 1998.

....7] Usually, neural networks are combined after training and are hence already quite perfect in 4 solving a classi cation or approximation problem on their own. The ensemble members are not trained in combination and the composition of the ensemble does not undergo an optimization process. In [17] neural networks are evolved and a subset of the nal population is combined afterwards. Di erent combination methods including averaging and majority voting are compared while a genetic algorithm is used to search for a near optimal ensemble composition. For genetic programming Zhang et al. ....

X. Yao and Y. Liu (1998) Making use of population information in evolutionary articial neural networks. IEEE Transactions on Systems, Man and Cybernetics, 28B(3):417-425.

No context found.

X. Yao, Y. Liu, Making use of population information in evolutionary artificial neural networks, in: IEEE Transactions on Systems, Man and Cybernetics, Part B: Cybernetics, Vol. 28, IEEE Press, 1998, pp. 417--425.

No context found.

X. Yao and Y. Liu. Making use of population information in evolutionary artificial neural networks. IEEE 388 Trans. on Systems, Man, and Cybernetics, Part B: Cybernetics, 28(3):417--425, 1998.

No context found.

X. Yao and Y. Liu, "Making use of population information in evolutionary artificial neural networks," IEEE Trans. on Systems, Man, and Cybernetics, Part B: Cybernetics, vol. 28, no. 3, pp. 417--425, June 1998.

....our future work. We expect incremental evolution to play a role in co evolutionary learning in further enhancing the generalization ability of evolved coalitions. 6 Conclusions Combining multiple players in a group can be a very effective way of designing a learning system that generalizes well [8, 9, 16], this paper further supports that this is a good approach. We have evolved coalitions consisting of a number of players in the IPD game. Each player in a coalition has a confidence value which specifies how well he she is in dealing with an opponent. The confidence value is not fixed. It is ....

X. Yao and Y. Liu, "Making use of population information in evolutionary artificial neural networks," IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics, 28(3):417-425, June 1998.

....significantly because there were good and poor networks for each case in the testing set and winner take all selected the best one. However it did not improved the independent training much TABLE XI COMPARISON AMONG CELS, EPNet [16] AN EVOLUTIONARY ENSEMBLE LEARNINIG ALGORITHM (Envo En RLS) [19], AND OTHERS [20] IN TERMS OF THE AVERAGE TESTING ERROR RATE FOR THE AUSTRALIAN CREDIT CARD ASSESSMENT PROBLEM. TER STANDS FOR TESTING ERROR RATE IN THE TABLE because the individual networks created by the independent training were all similar to each other. Table XI compares CELS s results ....

.... results with those produced by other neural and nonneural algorithms, where EPNet is an evolutionary system for designing neural networks [16] and Evo En RLS forms the final results by combining all the individuals in the last generation in EPNet based on the recursive least square algorithm [19]. The other algorithms represent the best 11 out of 23 algorithms tested in [20] Although CELS performed slightly worse than EPNet and Evo En RLS, it was significantly faster in terms of training time. CELS performed better than all other algorithms although they used ten fold cross validation. ....

X. Yao and Y. Liu, "Making use of population information in evolutionary artificial neural networks," IEEE Trans. Syst., Man, Cybern. B, vol. 28, pp. 417--425, June 1998.

No context found.

X. Yao and Y. Liu, "Making Use of Population Information in Evolutionary Neural Networks," IEEE Transactions on Systems, Man and Cybernetics - Part B: Cybernetics, vol. 28, no. 3, pp. 417--425, 1998.

No context found.

X. Yao and Y. Liu, Making use of population information in evolutionary artificial neural networks, IEEE Transactions on Systems, Man and Cybernetics - Part B: Cybernetics 28 (1998): 417-425.

No context found.

Xin Yao and Yong Liu. Making use of population information in evolutionary artificial neural networks. In IEEE Transactions on Systems, Man and Cybernetics, Part B: Cybernetics, volume 28, pages 417--425. IEEE Press, June 1998.

No context found.

X. Yao and Y. Liu. Making use of population information in evolutionary artificial neural networks. IEEE Transactions on Systems, Man and Cybernetics, Part B: Cybernetics, 28(3):417--425, June 1998.

No context found.

X. Yao and Y. Liu, Making use of population information in evolutionary artificial neural networks, IEEE Transactions on Systems, Man and Cybernetics - Part B: Cybernetics 28 (1998): 417-425.

No context found.

Xin Yao and Yong Liu. Making use of population information in evolutionary artificial neural networks. In IEEE Transactions on Systems, Man and Cybernetics, Part B: Cybernetics, volume 28, pages 417--425. IEEE Press, June 1998.

No context found.

X. Yao and Y. Liu, Making use of population information in evolutionary artificial neural networks, IEEE Trans. Systems, Man and Cybernetics - Part B: Cybernetics 28 (1998) 417-425.

First 50 documents

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