[Proceedings] 1991 IEEE International Joint Conference on Neural Networks

18-21 Nov. 1991

Filter Results

Displaying Results 1 - 25 of 444
  • Operational fault tolerance of the ADAM neural network system

    Publication Year: 1991, Page(s):1 - 6 vol.1
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (357 KB)

    The author investigates the fault tolerance of the Advanced Distributed Associative Memory (ADAM), focusing on its operational use. The effect of the reliability of recall of variously configured ADAM systems is examined by injecting faults individually, and also in various combinations since correlations between them will influence their overall effect on the system. Analysis of the results indic... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A general purpose neurocomputer

    Publication Year: 1991, Page(s):361 - 366 vol.1
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (252 KB)

    Presents a neural network, composed of linear units with threshold, as the CPU of a stored program MIMD architecture. The Caianiello formalism, is introduced as an aid to implement the arithmetic and control algorithms, needed for the smooth running of this general-purpose system. That is, in the neural net both the arithmetic and logic algorithms and the operating system have been implemented. Th... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Building up neuromimetic machines with LNeuro 1.0

    Publication Year: 1991, Page(s):602 - 607 vol.1
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (346 KB)

    The state of experiments on neural networks simulations on a parallel architecture is presented. The computing device, LNeuro 1.0, is based on an existing coarse-grain parallel framework (INMOS Transputers), improved with finer grain parallel abilities through VLSI modules. A digital architecture, scalable and flexible enough to be useful for simulating various kinds of networks and paradigms, was... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • 1991 IEEE International Joint Conference on Neural Networks (Cat. No.91CH3065-0)

    Publication Year: 1991
    Request permission for commercial reuse | PDF file iconPDF (47 KB)
    Freely Available from IEEE
  • Nonlinear neural field filters

    Publication Year: 1991, Page(s):1885 - 1890 vol.3
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (132 KB)

    Design, stability and implementation of nonlinear neural field filters are examined. The input and output of the neural field filters are vector fields. A neural transform is used to represent the input, output signals and the transfer function of the neural field filter. It is concluded that the Lyapunov conditions for such fields are taken care of by a novel extension of the Routh stability crit... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Approximations of mappings and application to translational invariant networks

    Publication Year: 1991, Page(s):2294 - 2298 vol.3
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (209 KB)

    The author studies the approximation of continuous mappings and dichotomies by one-hidden-layer networks, from a computational point of view. The approach is based on a new approximation method, specially designed for constructing small networks. Upper bounds are given on the size of these networks. These results are specialized to the case of transitional invariant networks, i.e., networks whose ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Associative memory design: Fokker-Planck formalism

    Publication Year: 1991, Page(s):2271 - 2276 vol.3
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (164 KB)

    A design methodology for associative memory (AM) using the Fokker-Planck formalism is proposed. This method allows AM to be designed for noisy conditions by taking into account the noise level to be tolerated by the AM as a design specification. The AM is viewed as a stochastic nonlinear dynamical system governed by a Fokker-Planck equation (FPE). Using eigenfunction expansion the spectrum of the ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • FastProp: a selective training algorithm for fast error propagation

    Publication Year: 1991, Page(s):2038 - 2043 vol.3
    Cited by:  Papers (3)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (312 KB)

    An improved backpropagation algorithm, called FastProp, for training a feedforward neural network is described. The unique feature of the algorithm is the selective training which is based on the instantaneous causal relationship between the input and output signals during the training process. The causal relationship is calculated based on the error backpropagated to the input layers. The accumul... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Learning temporary variable binding with dynamic links

    Publication Year: 1991, Page(s):2075 - 2079 vol.3
    Cited by:  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (260 KB)

    A novel gradient-based system for processing sequential time-varying inputs and outputs is described. With the method it is possible to train a system with time-varying inputs and outputs to use its dynamic links for temporarily binding variable contents to variable names as long as it is necessary for solving a particular task. Various learning methods for nonstationary environments are derived. ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A feedforward network of learning automata for pattern classification

    Publication Year: 1991, Page(s):2265 - 2270 vol.3
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (268 KB)

    A model made of units of teams of learning automata is developed for the three layer pattern classifier. The algorithm is approximated by an ordinary differential equation (ODE), using weak convergence methods. The pattern recognition problem is posed as a constrained maximization problem. It is shown that the zeros of the ODE correspond to points satisfying first order necessary conditions of the... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Networks of exponential neurons for multivariate function approximation

    Publication Year: 1991, Page(s):2305 - 2310 vol.3
    Cited by:  Papers (3)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (320 KB)

    A three-layer neural network, having a hidden layer of neurons with an exponential transfer function, capable of performing function approximation more accurately, and more economically, than a conventional multilayer perceptron (MLP) having neurons with a sigmoidal transfer function, is described. The network was trained by a variation of the standard backpropagation gradient-descent technique. T... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A modular approach to character recognition by neural networks

    Publication Year: 1991, Page(s):1309 - 1312 vol.2
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (96 KB)

    The main disadvantage of backpropagation neural networks has been the slow training rate in real-time situations where extensive training patterns and fairly extensive nets are the rule. Instead of training one huge net on the whole character set, the authors propose to implement parallel modular nets, each training on a very small character set. This approach can take full advantage of the distri... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Applying artificial neural networks to object and orientation recognition for robotic handling

    Publication Year: 1991, Page(s):1582 - 1587 vol.2
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (308 KB)

    Describes the application of artificial networks to recognition of objects and their orientation for the purpose of robotic handling of the objects. Three scenarios are considered: (1) two similar objects in five orientations; (2) two dissimilar objects in five orientations; and (3) three objects in three orientations. The orientations are identified by a rotation of the viewing position around on... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The effective dimension of the space of hidden units

    Publication Year: 1991, Page(s):2069 - 2074 vol.3
    Cited by:  Papers (20)  |  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (340 KB)

    The authors show how the effective number of parameters changes during backpropagation training by analyzing the eigenvalue spectra of the covariance matrix of hidden unit activations and of the matrix of weights between inputs and hidden units. They use the standard example of time series prediction of the sunspot series. The effective ranks of these matrices are equal to each other when a soluti... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Evolutionary stable and unstable strategies in neural networks

    Publication Year: 1991, Page(s):1448 - 1453 vol.2
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (304 KB)

    Neural networks simulate simple organisms that evolve in a two-dimensional environment (with selective reproduction and mutation) on the basis of a fitness criterion which rewards residing in favorable environmental zones and obtaining food. The evolution of two different populations of organisms is studied: organisms that are informed of the environmental zone they are in at any particular time a... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Micro-hardware implementation of a pattern recognition algorithm on a neuron-based multiprocessor system in a real-time environment

    Publication Year: 1991, Page(s):66 - 76 vol.1
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (432 KB)

    The authors present an experimental study on a system which implements a pattern recognition algorithm (W.C. Lin, K.S. Fu, 1965, 1966) with a set of artificial neuron elements emulated on a multi-microprocessor hardware system. The algorithm is believed to be good for parallel processing utilizing the concept of information content, or entropy. The system used for the implementation study consists... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Experiments in the application of neural networks to rotating machine fault diagnosis

    Publication Year: 1991, Page(s):769 - 774 vol.1
    Cited by:  Patents (4)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (368 KB)

    Reports a set of experiments performed to test the ability of a backpropagation network to classify the condition of an operating desktop fan based on its vibration signature. The trained network was used to detect and classify faults commonly occurring in industrial fans, i.e. impeller unbalance and cracked impeller blade. The discussion of the experimental results raises a number of issues relat... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Adaptive inertia compensation using a cerebellar model algorithm

    Publication Year: 1991, Page(s):2259 - 2264 vol.3
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (268 KB)

    A method for adaptively compensating for inertia in the control of robotic manipulators is described. The method uses an adaptive network which can be related to a number of contemporary models of the cerebellum. Simulations of a three-link planar manipulator show that the algorithm compensates for inertia during a movement without a priori knowledge. The fact that motions need not be repetitive g... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Some key factors in speaker recognition using neural networks approach

    Publication Year: 1991, Page(s):2752 - 2756 vol.3
    Cited by:  Papers (1)  |  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (188 KB)

    Some key factors influencing the performance of a speaker recognition system using a neural network approach are addressed. The selection of the input pattern affects the complexity and performance of the neural network. In the experiment, seven kinds of input patterns were selected. The complexity of the distribution of every kind of input pattern is different, which affects the training time and... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On the bound of the approximation capacity of multi-layer neural network

    Publication Year: 1991, Page(s):2299 - 2304 vol.3
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (148 KB)

    The authors deal with the approximation capacity of a multilayer neural network (MLNN) and present a strict mathematical expression for the bound of the approximation ability of a single input-output MLNN with hidden units of sigmoid activation functions. The influence of the number of hidden units, connections, and the threshold on the approximation accuracy is also discussed. The relationship be... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An investigation into neural net control systems with integral action

    Publication Year: 1991, Page(s):1053 - 1058 vol.2
    Cited by:  Papers (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (232 KB)

    The authors investigate the incorporation of integral action for a neural net control system applicable to a class of nonlinear dynamical systems. The properties and performances of INC/NN (inverse nonlinear control/neural network) controllers are discussed and studied. Simulation results on applying the proposed INC/NN controller to a position control system with inherent nonlinearities are prese... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A cognitively-based neural network for determining paragraph coherence

    Publication Year: 1991, Page(s):1303 - 1308 vol.2
    Cited by:  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (272 KB)

    The authors report on an effort in artificial neural network (ANN) technology to use content-independent elements of prose as predictors of paragraph logic structures. They intend to embed the trained network in an intelligent tutor to teach writing skills. An attempt is made to find patterns in the nonambiguous lexical and syntactic features if discourse that predict the semantic/cognitive level ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A multilayered neural net controller for servo systems

    Publication Year: 1991, Page(s):1577 - 1581 vol.2
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (256 KB)

    The authors investigate the possibility of adding a multilayered feedforward neural network controller to an existing servomotor controller to make it an intelligent adaptive controller. The use of the existing controller guarantees coarse learning and thus provides better generalization and correction capabilities. Several learning algorithms are proposed to properly correct the motor inputs unde... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Improving the training speed of three-layer feedforward neural nets by optimal estimation of the initial weights

    Publication Year: 1991, Page(s):2063 - 2068 vol.3
    Cited by:  Papers (15)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (244 KB)

    The authors formulate the training problem for three-layer feedforward neural nets based on the well known linear algebra of D. Rumelhart et al. (1986). Then, they develop two estimation algorithms, called the forward estimation algorithm and the recurrent estimation algorithm, to estimate the initial weights. The basic idea is to set the initial weights space as close as possible to a global mini... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Control of a robotic manipulating arm by a neural network simulation of the human cerebral and cerebellar cortical processes

    Publication Year: 1991, Page(s):1444 - 1447 vol.2
    Cited by:  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (188 KB)

    The authors propose a system commanding a robotic manipulating arm under visual control, based on brain modeling. In this model, the movement command is learned by a network which links two subsystems together: a cerebral subsystem which can learn a goal, and a second subsystem responsible for quantitative adjustments and coordination. Those two subsystems are complementary because each subsystem ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.