A. P. Wieland. Evolving neural network controllers for unstable systems. In IEEE International Joint Conference on Neural Networks, pages II-667 -- II-673, IEEE Press, Seattle, WA, 1990.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....to various evolutionary search algorithms. 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, ....

A. Wieland. Evolving neural network controllers for unstable systems. In Proceedings of the International Joint Conference on Neural Networks, pages 667--673, 1991.

....often termed chromosomes. Some of the early work in evolving ANN connection weights followed this approach Fig. 3. a) An ANN with connection weights shown. b) A binary representation of the weights, assuming that each weight is represented by four bits. 24] 26] 28] 37] 38] [41], 52] 53] In such a representation scheme, each connection weight is represented by a number of bits with certain length. An ANN is encoded by concatenation of all the connection weights of the network in the chromosome. A heuristic concerning the order of the concatenation is to put ....

....and nondifferentiable surface. It does not depend on gradient information of the error (or fitness) function and thus is particularly appealing when this information is unavailable or very costly to obtain or estimate. For example, the evolutionary approach has been used to train recurrent ANN s [41], 60] 65] 100] 102] 103] 106] 117] 126] 128] higher order ANN s [52] 53] and fuzzy ANN s [76] 77] 129] 130] Moreover, the same EA can be used to train many different networks regardless of whether they are feedforward, recurrent, or higher order ANN s. The general ....

A. P. Wieland, "Evolving neural network controllers for unstable systems," in Proc. 1991 IEEE Int. Joint Conf. Neural Networks (IJCNN'91 Seattle), vol. 2, pp. 667--673.

....will explain how each ablation was performed and interpret the results. B. No growth Ablation In order to make no growth NEAT comparable to fixedtopology NE, it was allowed to start with a fully connected hidden layer of 10 hidden units, the same number as in past fixed topology NE experiments [15, 19]. Without growth, the system was only able to use weight differences to speciate the population. Given 1,000 generations to find a solution, the ablated system could only find a solution 20 of the time When it did find a solution, it took 8.5 times more evaluations than full NEAT. Clearly, ....

....with new structure. Thus, when a species in NEAT is on a local optimum, it is possible that by adding a new connection, a new dimension of freedom may open up, leading to a path away from the local optimum. A parallel can be drawn between structure evolution in NEAT and incremental evolution [5, 19]. Incremental evolution is a method used to train a system to solve harder tasks than it normally could by training it on incrementally more challenging tasks. The idea is that NE is likely to get stuck on a local optimum when attempting to solve the harder task directly. However, after solving ....

A. P. Wieland. Evolving neural network controllers for unstable systems. In Proceedings of the International Joint Conference on Neural Networks (Seattle, WA), pages 667--673. Piscataway, NJ: IEEE, 1991.

....comparison have been shown superior to reinforcement learning methods elsewhere (Moriarty and Miikkulainen 1996) Thus, the question here is whether evolving structure can lead to greater NE performance. 4.3. 1 Pole Balancing Comparisons We set up the pole balancing experiments as described by Wieland (1991) and Gomez and Miikkulainen (1999) The Runge Kutta fourth order method was used to implement the dynamics of the system, with a step size of 0.01s. All state variables were scaled to [ 1:0; 1:0] before being fed to the network. Networks output a force every 0.02 seconds between [ 10; 10]N . The ....

....307,200 150 2048 Conventional NE 80,000 800 100 SANE 12,600 63 200 ESP 3,800 19 200 NEAT 3,600 24 150 Table 1: Double Pole Balancing with Velocity Information. Evolutionary programming results were obtained by Saravanan and Fogel (1995) Conventional neuroevolution data was reported by Wieland (1991). SANE and ESP results were reported by Gomez and Miikkulainen (1999) NEAT results are averaged over 120 experiments. All other results are averages over 50 runs. The standard deviation for the NEAT evaluations is 2,704 evaluations. Although standard deviations for other methods were not ....

[Article contains additional citation context not shown here]

Wieland, A. (1991). Evolving neural network controllers for unstable systems. In Proceedings of the International Joint Conference on Neural Networks (Seattle, WA), 667--673. Piscataway, NJ: IEEE.

....are commented. I. Introduction For years, evolutionary methods have been used in combination with neural networks. In older approaches, the network structure remains fixed and the connection weights are just evolved, with a genetic algorithm e.g. performing a kind of evolutionary learning ( 1] [2]) There is a strong limitation to generalize as the number of neurons and the structure of the network are set with an a priori knowledge on the problem. Moreover, there is no general method for determining the structure of a neural network fitted to a particular problem. There are only problem ....

....problem. The adaline model was trained to balance poles using Widrow Hoff LMS algorithm ( 11] Reinforcement learning methods have also addressed this problem ( 12] The cart pole system is still a standard research problem for neural networks. There are two recent examples, drawn from Wieland ([2]) and Whitley and al. 13] where a Genetic Algorithm is used to evolve Neural Networks to control the cart pole system. The Wieland s method optimizes the weights of a fixed fully recurrent Neural Network, using a Genetic Algorithm. This method leads to very interesting results but is not ....

A. Wieland, "Evolving neural network controllers for unstable systems", in International Joint Conference on Neural Networks, 1991, pp. 667--673.

....evolve neural controllers able to balance an inverted pendulum mounted on a cart and centering the cart simultaneously. This is decribed in section 3. Besides conventional control techniques, there have been many successfull applications of neural networks to this problem [2] 4] 7] 11] 6] [13], which can be compared with solutions obtained by the presented evolutionary algorithm. Using continuous neurons for the controllers, di erent from many other results, our approach does not make use of quantization, neither of the physical phase space variables nor of internal parameters, like ....

Wieland, A. P. (1991). Evolving neural network controllers for unstable systems. In: International Joint Conference on Neural Networks, Seattle, USA, July1991. Proccedings Vol.2. Seattle: IEEE Service Center, 1991.

....Kevin W. 944] White, David W. 27] White, D. 653] Whitehead, B. A. 418] Whitehead, Bruce A. 419] Whitfort, T. 514] Whitley, Darrell L. 341] Whitley, Darrell, 132, 303, 314, 908, 945, 946, 947, 948, 949, 950, 951, 952, 953, 954, 955, 956, 957, 958, 959] Wieland, Alexis P. [960] Wienholt, Willfried, 961, 962] Wilke, Peter, 963, 964] Wilke, P. 229] Williams, Bryn V. 631] Williams, G. J. 542] Williams, Tom, 220] Williamson, A. G. 288] Wilson, Stewart W. 902] Winfield, A. 378] Winkler, David A. 495] Wise, B. M. 149] Wong, F. 35] Woo, ....

.... deterministic, 466] dynamic, 295] neural Darwinism, 556] neural netoworks, 243] neural network, 539] control, 280, 515] design, 444] fuzzy, 423, 477] rule extraction, 464] signal processing, 287] structure selection, 172] training, 354] wavelet, 281] neural networks, [619, 665, 666, 914, 939, 599, 600, 650, 651, 727, 783, 799, 945, 946, 601, 606, 613, 642, 649, 656, 657, 711, 728, 729, 854, 909, 947, 948, 949, 950, 951, 952, 970, 602, 607, 610, 615, 634, 637, 645, 652, 663, 670, 678, 679, 680, 681, 682, 683, 684, 685, 714, 721, 730, 756, 757, 762, 764, 770, 781, 782, 785, 786, 787, 800, 810, 815, 834, 869, 874, 875, 879, 885, 902, 904, 907, 910, 920, 923, 940, 953, 954, 971, 973, 974, 975, 976, 977, 608, 618, 620, 622, 628, 630, 644, 654, 660, 661, 664, 674, 686, 687, 688, 693, 712, 713, 731, 735, 758, 763, 793, 796, 806, 813, 817, 841, 857, 859, 860, 861, 864, 886, 889, 890, 911, 913, 915, 921, 936, 942, 955, 956, 960, 972, 978, 979, 980, 981, 982, 592, 594, 595, 605, 611, 614, 616, 621, 635, 636, 638, 639, 643, 658, 669, 671, 672, 673, 689, 690, 691, 696, 700, 716, 717, 719, 720, 722, 723, 724, 725, 726, 733, 734, 736, 737, 738, 739, 740, 741, 766, 767, 773, 774, 778, 784, 790, 791, 795, 797, 801, 804, 809, 814, 818, 819, 821, 822, 823, 824, 829, 835, 836, 837, 838, 845, 846, 858, 871, 876, 880, 881, 900, 906, 908, 917, 919, 922, 925, 926, 941, 957, 958, 965, 967, 968, 983, 984, 593, 596, 597, 598, 609, 612, 617, 629, 631, 632, 633, 640, 641, 646, 653, 659, 662, 668, 675, 677, 692, 694, 698, 701, 703, 704, 705, 706, 709, 742, 743, 744, 745, 746, 747, 748, 749, 750, 751, 752, 753, 761, 765, 768, 769, 771, 775, 776, 780, 788, 794, 802, 803, 805, 807, 808, 811, 812, 825, 827, 832, 833, 842, 844, 847, 848, 849, 850, 856, 862, 863, 865, 866, 868, 870, 872, 873, 877, 884, 887, 892, 893, 894, 895, 896, 901, 905, 918, 927, 929, 930, 931, 933, 934, 935, 937, 938, 943, 944, 961, 963, 966, 985, 986, 987, 988, 13, 14, 18, 26, 38, 40, 43, 63, 75, 84, 86, 88, 91, 96, 99, 103, 106, 108, 111, 112, 113, 114, 118, 121, 124, 125, 128, 129, 134, 137, 138, 139, 140, 146, 147, 151, 153, 156, 157, 166, 167, 169, 180, 190, 191, 192, 193, 196, 199, 200, 209, 215, 217, 222, 226, 227, 230, 236, 238, 241, 242, 244, 254, 258, 260, 264, 266, 268, 269, 271, 272, 277, 279, 283, 292, 295, 296, 297, 303, 305, 318, 321, 322, 323, 327, 333, 338, 349, 353, 366, 372, 374, 375, 378, 381, 388, 390, 398, 400, 403, 404, 407, 408, 411, 415, 427, 440, 441, 448, 453, 454, 456, 458, 459, 463, 473, 360, 475, 478, 481, 487, 489, 490, 491, 492, 493, 495, 499, 504, 505, 508, 512, 523, 525, 526, 530, 531, 533, 542, 556, 562, 566, 583] neural networks age, 206] analysis, 362] architecture, 41, 184] associative memory, 570] back propagation, 397] back propagation, 502] backpropagation, 64, 69, 115, 351, 447] Baldwin effect, 155] Bayesian, 174, 339, 527] binary logic, 565] biological, 779] Boltzmann, 70] ....

[Article contains additional citation context not shown here]

Alexis P. Wieland. Evolving neural network controllers for unstable systems. In 1991 International Joint Conference on Neural Networks - IJCNN 91, volume II, pages 667--673, Seattle, WA, 8.-14. July 1991. IEEE, New York. ga:Wieland91.

....the system status. In real applications, practitioners may find difficulty in acquiring system information such as the cart velocity and the angular velocity. Therefore, it is important to obtain a neurocontroller with as few input variables as possible. This problem can be tackled, as in [35], by using an RNN in which h and are the only inputs. In our experiments, an RNN with two inputs and six processing nodes (one of them also serves as an output node) as shown in Fig. 5a, was used to balance the pendulum. 4 At each time step, and h were applied as inputs to the RNN. If the ....

A. Wieland. Evolving neural network controllers for unstable systems. In Proceedings of the International Joint Conference on Neural Networks, pages 667--673, 1991.

....a training set of inputoutput pairs does not exist. Rather, the neural network is applied to a problem and the performance of the network is used to supply a reinforcement signal. One particularly interesting example of this kind of work is the neurocontrol pole balancing experiments of Alexis Wieland (1990;1991). Using a genetic algorithm Wieland trained fully recurrent neural networks to balance a pole fixed to a cart moving on a finite track using only the pole angle and cart position as inputs. Typically, velocity information is provided for this problem, but in one variant of this problem a fully ....

Wieland, A. (1991). Evolving Neural Network Controllers for Unstable Systems. In International Joint Conference on Neural Networks, Seattle, pages 667--673.

....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 ....

....Figure 2: A typical cycle of the evolution of connection weights. 2.1 Binary Representation The canonical GA [13, 14] has always used binary strings to encode alternative solutions, often termed chromosomes. Some of the early work in evolving ANN connection weights followed this approach [24, 26, 28, 37, 38, 41, 52, 53]. In such a representation scheme, each connection weight is represented by a number of bits with certain length. An ANN is encoded by concatenation of all the connection weights of the network in the chromosome. A heuristic concerning the order of the concatenation is to put connection weights to ....

[Article contains additional citation context not shown here]

A. P. Wieland, "Evolving neural network controllers for unstable systems," in Proc. of 1991 IEEE International Joint Conference on Neural Networks (IJCNN'91 Seattle), vol. 2, pp. 667--673, IEEE Press, New York, NY, 1991.

....in [Whitley, 1993] to apply GANN systems to control problems that rely on reinforcement learning. Since no target output is given, back propagation cannot be used as a training method. In general, control problems demand the training of a real valued function. In case of the pole balancing problem [Wieland, 1991], the input values represent state information such as angle and velocity of the pole. The output values represent action information such as the force that has to be applied to the pole to keep it in balance. So, basically we can model each reinforcement control task with the training of a real ....

: Alexis P. Wieland: "Evolving Neural Network Controllers for Unstable Systems", in: International Joint Conference on Neural Networks, Vol. II, pp. 667-673, IEEE.

....enough for modern reinforcement learning methods and more difficult variants need to be found. One particularly difficult variant is a cart with two poles of different lengths that have to be balanced simultaneously. The 2 pole problem can be solved with direct neuro evolution methods (Wieland, 1990, 1991), but is still hard enough to pose a challenge to even the more recent composite neuro evolution methods, such as SANE and Cellular Encoding (Whitley et al. 1995; Gruau et al. 1996a,b) One particularly difficult variant is a cart with two poles of different lengths that have to be balanced ....

....methods, such as SANE and Cellular Encoding (Whitley et al. 1995; Gruau et al. 1996a,b) One particularly difficult variant is a cart with two poles of different lengths that have to be balanced simultaneously. This problem has been first solved successfully with direct neuro evolution methods (Wieland, 1990, 1991), but composite neuro evolution methods like Cellular Encoding (Gruau et al. 1996a) and symbiotic evolution (SANE) have proved to be far superior to the direct approaches. This paper uses the SANE neuroevolution algorithm as a starting point, because it appears to be currently among the strongest ....

[Article contains additional citation context not shown here]

Wieland, A. (1991). Evolving neural network controllers for unstable systems. In Proc. IJCNN (Seattle, WA), vol. II, 667--673. Piscataway, NJ: IEEE.

....evolve neural networks using genetic algorithms) for example, often find solutions in the initial random population [Moriarty and Miikkulainen, 1996; Gomez and Miikkulainen, 1997] In response to this need for a new benchmark, the basic pole balancing task has been extended in a variety of ways. Wieland[1991] presented several variations to the standard single pole task that can be grouped into two categories: 1) modifications to the mechanical system itself such as adding a second pole either next to or on top of the other. 2) restricting the amount of state information that is given to the ....

....lengths. When the poles are very close in length, solutions to this system cannot be evolved directly by current methods. In order to control the system under these conditions, shaping (or incremental learning) techniques can be employed that increase the length of the shorter pole very gradually [Wieland, 1991; Saravanan and Fogel, 1995] This kind of approach is effective but can be extremely slow due to the limitations of the underlying evolutionary search method many generations are required to recover from minute changes to the environment. Using an incremental approach in conjunction with a ....

[Article contains additional citation context not shown here]

Alexis Wieland. Evolving neural network controllers for unstable systems. In Proceedings of the International Joint Conference on Neural Networks (Seattle, WA), volume II, pages 667--673, Piscataway, NJ, 1991. IEEE.

....in [Whitley, 1993] to apply GANN systems to control problems that rely on reinforcement learning. Since no target output is given, back propagation cannot be used as a training method. In general, control problems demand the training of a real valued function. In case of the pole balancing problem [Wieland, 1991], the input values represent state information such as angle and velocity of the pole. The output values represent action information such as the force that has to be applied to the pole to keep it in balance. So, basically we can model each reinforcement control task with the training of a real ....

: Alexis P. Wieland: "Evolving Neural Network Controllers for Unstable Systems ", in: International Joint Conference on Neural Networks, Vol. II, pp. 667-673, IEEE.

....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 ....

....weights. Reprinted with permission from Ref. 1. X. Yao: Evolutionary Artificial Neural Networks 9 2. 1 Binary Representation Since the binary representation has been shown to be beneficial in GA s search [4, 5] one way to represent connection weights is to encode them in binary strings [29, 30, 32, 41, 42, 45, 56, 57]. In such a representation scheme, each connection weight is represented by a number of binary bits with certain length. An EANN is represented by concatenation of all the connection weights in the network. A heuristic concerning the order of the concatenation is to put connection weights to the ....

[Article contains additional citation context not shown here]

A. P. Wieland. Evolving neural network controllers for unstable systems. In Proc. of 1991 IEEE International Joint Conference on Neural Networks (IJCNN'91 Seattle), volume 2, pages 667--673. IEEE Press, New York, NY, 1991. X. Yao: Evolutionary Artificial Neural Networks 48

....assumes that components of all parent representations may be freely exchanged without inhibiting the search process. Various combinations of GAs and connectionist networks have been investigated. Much research concentrates on the acquisition of parameters for a fixed network architecture (e.g. [18 21]) Other work allows a variable topology, but disassociates structure acquisition from acquisition of weight values by interweaving a GA search for network topology with a traditional parametric training algorithm (e.g. backpropagation) over weights (e.g. 22, 23] Some studies attempt to ....

A. P. Wieland, "Evolving neural network controllers for unstable systems," In IEEE International Joint Conference on Neural Networks, IEEE Press, Seattle, WA, pp. II-667 -- II-673, 1990.

....(NE) systems (i.e. systems that evolve neural networks using genetic algorithms) for example, often find solutions in the initial random population [5, 6] In response to this need for a new benchmark, a variety of ways to extend the basic pole balancing task have been suggested. Wieland [7] presented a series of increasingly difficult variations on the standard pole balancing task This research was supported in part by National Science Foundation under grant #IRI9504317. Task Environment Figure 1: The Enforced Sub Populations Method (ESP) The population of neurons is segregated ....

.... variables: the angle of the pole from vertical in the x and y directions ( x , y ) the angular velocities of the pole ( x , y ) the position of the cart in the plane (x,y) and the velocity of the cart ( x, y) The equations of motion for this system are an extension of those found in [7] for the single pole problem with one equation for each principal axis. In all of the experiments we performed the networks were only provided with x, y, x , and y , and the network itself had to determine the Figure 2: The 2 degree of freedom pole balancing system. The figure shows a ....

A. Wieland. Evolving neural network controllers for unstable systems. In Proceedings of the International Joint Conference on Neural Networks (Seattle, WA), volume II, pages 667--673, Piscataway, NJ, 1991. IEEE.

....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 ....

....evolution of connection weights is shown in Figure 1. X. Yao: Evolutionary Artificial Neural Networks 5 2. 1 Binary Representation Since the binary representation has been shown to be beneficial in GA s search [11, 12] one way to represent connection weights is to encode them in binary strings [27, 29, 31, 40, 41, 44, 55]. In such a representation scheme, each connection weight is represented by a number of binary bits with certain length. For example, Whitley et al. 27, 29] used 8 bits to represent each connection weight, which ranges between Gamma127 and 127, in their experiments with XOR and adder ....

[Article contains additional citation context not shown here]

A. P. Wieland. Evolving neural network controllers for unstable systems. In Proc. of 1991 IEEE International Joint Conference on Neural Networks (IJCNN'91 Seattle), volume 2, pages 667--673. IEEE Press, New York, NY, 1991.

....Singh 1992) Typically, in these approaches the complex task is broken into simpler components or subtasks that are each learned by separate systems (e.g. GAs or rule bases) and then combined to achieve the goal task. In contrast, in incremental evolution as proposed in this paper and also used by Wieland (1990, 1991) and Saravanan and Fogel (1995) a single system learns a succession of tasks. Such an adaptation process is similar to continual (or lifelong) learning (Elman 1991; Ring 1994; Thrun 1996) and motivated by learning in real life. If, for instance, the goal task is that of driving a Formula 1 race ....

....allow the force to be continuous within a specified range. Second, the controller may be provided with only x and , so that it has to use recurrent connections to compute the derivatives x and in order to balance the pole. Using a conventional NE method on networks with six recurrent units, Wieland (1990, 1991) was able to evolve a controller for this problem in an average of 10 generations. ESP solved the same task in as many generations on average (over 10 simulations) but tested less than half as many networks (200 instead of 512 per generation) An even more challenging problem is to place a second ....

Wieland, A. (1991). Evolving neural network controllers for unstable systems. In Proceedings of the International Joint Conference on Neural Networks (Seattle, WA), vol. II, 667--673.

....different transfer functions like tanh or the strictly positive sigmoide (1 e Gammax ) Gamma1 , are described in section 3. Besides conventional control techniques, there have been many successful applications of neural networks to the pole balancing problem e.g. 2] 4] 6] 19] [21]. Using continuous neurons for the controllers, in contrast to many other results, our approach does not make use of quantization either of the physical phase space variables or of internal network parameters, such as weights and bias terms or output values. Compared with other neural network ....

Wieland, A. P. (1991). Evolving neural network controllers for unstable systems. In: International Joint Conference on Neural Networks, Seattle, USA, July1991. Proccedings Vol.2. Seattle: IEEE Service Center, 1991.

No context found.

A. P. Wieland. Evolving neural network controllers for unstable systems. In IEEE International Joint Conference on Neural Networks, pages II-667 -- II-673, IEEE Press, Seattle, WA, 1990.

No context found.

Wieland, A. (1991), Evolving neural network controllers for unstable systems, in `Proceedings of the International Joint Conference on Neural Networks , IEEE Press, pp. 667--673.

No context found.

A. Wieland. Evolving neural network controllers for unstable systems. In International Joint Conference on Neural Networks, 1991.

No context found.

A. Wieland. Evolving neural network controllers for unstable systems. In Interna- 22 Artificial Neurogenesis Manuscripts : : : tional Joint Conference on Neural Networks, pages 667--673, 1991.

No context found.

Alexis P. Wieland. Evolving Neural Network Controllers for Unstable Systems. IJCNN-91, II:667--673, 1991.

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