IEEE International Conference on Neural Networks

28 March-1 April 1993

Filter Results

Displaying Results 1 - 25 of 341
  • A distributed adaptive control system for a quadruped mobile robot

    Publication Year: 1993, Page(s):144 - 149 vol.1
    Cited by:  Papers (4)  |  Patents (2)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (506 KB)

    A method by which reinforcement learning can be combined into a behavior based control system is presented. Behaviors which are impossible or impractical to embed as predetermined responses are learned through self-exploration and self-organization using a temporal difference reinforcement learning technique. This results in what is referred to as a distributed adaptive control system (DACS), whic... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A cortical structure for real world image processing

    Publication Year: 1993, Page(s):138 - 143 vol.1
    Cited by:  Papers (1)  |  Patents (1)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (585 KB)

    Neural architecture as found in the mammalian visual cortex is used for visual processing of real world camera images. The neural architecture used does not refer to classical neural nets but to more global characteristics such as the typical receptive field characteristics, two-dimensional cortical structure, local operations and topographic arrangement of cells. The self-organization algorithm a... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A neural network based edge detector

    Publication Year: 1993, Page(s):132 - 137 vol.1
    Cited by:  Papers (7)  |  Patents (1)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (674 KB)

    An approach to the edge detection problem based on the nonlinear mapping and generalization capabilities of multilayer feed forward neural networks is proposed. The task of edge detection is broken into two parts, i.e., mapping typical gray levels in primitive small image blocks (e.g., 3*3 windows) to their corresponding most likely edge patterns using a simple neural network, and combining this l... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Stereo correspondence through multiple constraint neural networks

    Publication Year: 1993, Page(s):126 - 131 vol.1
    Cited by:  Papers (2)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (678 KB)

    A fast and robust artificial neural network algorithm for solving the stereo correspondence problem in binocular vision is described. The stereo correspondence is modeled as a cost minimization problem where the cost is the value of the matching function between the edge pixels along the same epipolar line. A multiple-constraint energy minimization neural network is implemented for this matching p... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Analog neural networks solve ambiguity problems in medium PRF radar systems

    Publication Year: 1993, Page(s):120 - 125 vol.1
    Cited by:  Papers (1)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (554 KB)

    Medium pulse-repetition frequency (PRF) radars combine the features of high PRF radars and low PRF radars. Both range and Doppler (range rate) ambiguities exist in such radars. It is demonstrated that the ambiguity problems in medium PRF radars can be solved efficiently using the neural network approach. A multilayer feedforward network is designed to solve the ambiguity problems. Both the simulat... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An analog VLSI model of central pattern generation in the medicinal leech

    Publication Year: 1993, Page(s):116 - 119 vol.1
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (324 KB)

    The design and construction of a working silicon model of the neural system responsible for swimming behaviors of the medicinal leech (Hirundo medicinalis) are presented. The silicon model spans three levels of organization in the leech nervous system (neuron, ganglion, system) and represents one of the first comprehensive models of leech swimming. The circuit employs biophysically motivated analo... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The effects of analog hardware properties on backpropagation networks with on-chip learning

    Publication Year: 1993, Page(s):110 - 115 vol.1
    Cited by:  Papers (7)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (595 KB)

    Results of simulations performed assuming both forward and backward computation done on-chip using analog components are presented. Aspects of analog hardware studied are component variability (variability in multiplier gains and zero offsets), limited voltage ranges, and components (multipliers) that only approximate the computations in the backpropagation algorithm. It is shown that backpropagat... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Force directed self-organizing map and its application to VLSI cell placement

    Publication Year: 1993, Page(s):103 - 109 vol.1
    Cited by:  Papers (9)  |  Patents (1)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (500 KB)

    A self-organizing map is described. It is called the force directed self-organizing map (FDSOM), and can be used in VLSI cell placement with various constraints on connection and dimension, such that the total wire length and area of the resulting placement are minimized. This procedure combines ideas from force-directed relaxation and the self-organization algorithm proposed by Kohonen. It is par... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Solving inverse problems in nonlinear PDEs by recurrent neural networks

    Publication Year: 1993, Page(s):99 - 102 vol.1
    Cited by:  Papers (1)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (267 KB)

    A neural network approach for solving inverse problems in nonlinear partial differential equations (PDEs) is proposed, and a computer simulation based on this approach is described. The network is designed based on the differential difference equation (DDE) approximating the PDE. The network is trained so that its output and the known boundary values of connection weights and thresholds represent ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Classification of Parkinson rating-scale-data using a selforganising neural net

    Publication Year: 1993, Page(s):93 - 98 vol.1
    Cited by:  Papers (4)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (614 KB)

    An application of a self-organizing neural net of Kohonen type to the data of 666 de-novo Parkinsonian patients of a multicenter study is presented. The data to be learned are the ten items of the Webster rating scale and one additional item with four stages, following the classification by Hoehn and Yahr. Multivariate linear statistical methods are applied to the data, yielding linear models, whi... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Solving the serializability problem by a connectionist machine

    Publication Year: 1993, Page(s):87 - 92 vol.1
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (493 KB)

    The serializability problem, which is NP-complete, is to decide on the existence of a total order consistent with a given ordering constraint. A connectionist machine is proposed for finding an almost sure solution to the serializability problem such that a feasible final configuration implies that the set is serializable with the given ordering constraint. An infeasible final configuration does n... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Solving search problems with subgoals using an artificial neural network

    Publication Year: 1993, Page(s):81 - 86 vol.1
    Cited by:  Papers (1)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (461 KB)

    Search problems with a series of subgoals can be solved using symbolic search algorithms. A method is proposed to use a neural network to perform this type of search by translating the serial and temporal resolution path into a spatial and parallel constraint structure using both state units and constraint units. A network is designed for the Missionaries and Cannibals Problem to illustrate the me... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Dynamics of synaptic transfer in living and simulated neurons

    Publication Year: 1993, Page(s):75 - 80 vol.1
    Cited by:  Papers (5)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (550 KB)

    A comparison of the responses of an ionic-permeability-based neural model to periodic inhibitory driving with that of a living preparation is summarized. Duplication of most neuron response types is excellent. Simulation results lead to insights into neuron activities that are verified by examination of the living data.<<ETX>> View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A spatio-temporal neural network model of saccade generation

    Publication Year: 1993, Page(s):70 - 74 vol.1
    Cited by:  Papers (5)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (371 KB)

    A spatio-temporal neural network model of the superior colliculus is presented. The model uses lateral excitatory and inhibitory connections to help control both the dynamic and static behavior of saccadic eye movements. In simulations, the model succeeds in replicating accurate saccades of a variety of sizes. Simulation results and the model's relation to neurophysiological findings are discussed... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Temporal competition as an optimal parallel processing of the cerebrohypothalamic system

    Publication Year: 1993, Page(s):64 - 69 vol.1
    Cited by:  Papers (3)  |  Patents (1)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (414 KB)

    Cortical processing which makes responses in a few hundred milliseconds using neurons firing at less than 50 Hz is analyzed. A mathematical model is presented to show the cortical computational architecture using a few spikes of each neuron. The cortex is represented by a sequence of areas. Each area consists of cortical columns. The columns include pyramidal cells and interneurons laterally inhib... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Synaptic coding of periodically modulated spike trains

    Publication Year: 1993, Page(s):58 - 63 vol.1
    Cited by:  Papers (5)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (479 KB)

    The authors describe how periodically modulated pre-synaptic trains influence postsynaptic discharges in pacemaker neurons. Experiments are conducted on inhibitory synapses in crayfish. Descriptions are based upon point-process and dynamical systems theories. Pre- and postsynaptic discharges differ clearly. Local marked distortions are due to similar trends, special patterns, asymmetric sensitivit... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A hybrid technique to enhance the performance of recurrent neural networks for time series prediction

    Publication Year: 1993, Page(s):52 - 57 vol.1
    Cited by:  Papers (2)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (487 KB)

    The recurrent neural networks trained by the real time recurrent learning (RTRL) algorithm is used for time series prediction. When there is a strong nonlinear relationship connecting the adjacent samples of the time series which the network is trying to predict, the prediction performance of the network deteriorates. A scheme is proposed to overcome this drawback. This scheme incorporates cascade... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On design and evaluation of tapped-delay neural network architectures

    Publication Year: 1993, Page(s):46 - 51 vol.1
    Cited by:  Papers (38)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (457 KB)

    Pruning and evaluation of tapped-delay neural networks for the sunspot benchmark series are addressed. It is shown that the generalization ability of the networks can be improved by pruning using the optimal brain damage method of Le Cun, Denker and Solla. A stop criterion for the pruning algorithm is formulated using a modified version of Akaike's final prediction error estimate. With the propose... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Hopf bifurcation and Hopf hopping in recurrent nets

    Publication Year: 1993, Page(s):39 - 45 vol.1
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (519 KB)

    Some aspects of the learning dynamics of recurrent neural networks are discussed. It is shown that as a two-unit fully recurrent network is trained to oscillate, the learning process brings the network to a point where a small change in any one of the weights can push the network through a Hopf bifurcation to create stable oscillation. As learning continues, the network indeed bifurcates to create... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Self-clustering recurrent networks

    Publication Year: 1993, Page(s):33 - 38 vol.1
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (517 KB)

    It is shown, based on empirical analyses, that second-order recurrent neural networks which are trained to learn finite state automata (FSAs) tend to form discrete clusters as the state representation in the hidden unit activation space. This observation is used to define self-clustering networks which automatically extract discrete state machines from the learning network. To address the problem ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Parallel consensual neural networks

    Publication Year: 1993, Page(s):27 - 32 vol.1
    Cited by:  Papers (6)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (576 KB)

    A neural network architecture is proposed and applied in classification of remote sensing/geographic data from multiple sources. The architecture is called the parallel consensual neural network, and its relation to hierarchical and ensemble neural networks is discussed. The parallel consensual neural network architecture is based on statistical consensus theory. The input data are transformed sev... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Back-propagation algorithm with controlled oscillation of weights

    Publication Year: 1993, Page(s):21 - 26 vol.1
    Cited by:  Papers (2)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (524 KB)

    Error back propagation (EBP) is a training algorithm for feedforward artificial neural networks (FFANNs). Certain simple but effective modifications to the EBP algorithm to improve its convergence are proposed. The methods watch for oscillation of weights as the training algorithm proceeds. When such an oscillation is observed, the learning rate for only that weight is temporarily reduced. Effecti... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A universal structure for artificial neural networks

    Publication Year: 1993, Page(s):15 - 20 vol.1
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (522 KB)

    An approximation procedure, named successive linearization, is introduced for unified implementation of a large class of neural networks. A nonlinear neural model with dynamic sensitivity is presented. It is modular and has rapid learning schemes. Arbitrary nonlinear functions with memory which are commonly used for modeling dynamical systems, as well as static nonlinear classification boundaries,... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Multiple networks for function learning

    Publication Year: 1993, Page(s):9 - 14 vol.1
    Cited by:  Papers (40)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (463 KB)

    In the case of learning a mapping, it is proposed to build several possible models instead of one, train them all independently on the same task and take a vote over their responses. These networks will converge to different solutions due to using different models, different parameter set sizes or any other factor related to training. Two training methods are used, i.e., grow and learn (GAL), a me... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Learning algorithms for adaptive processing and control

    Publication Year: 1993, Page(s):1 - 8 vol.1
    Cited by:  Papers (2)
    Request permission for reuse | Click to expandAbstract | PDF file iconPDF (680 KB)

    Linear and nonlinear adaptive filtering algorithms are described, along with applications to signal processing and control problems. Specific topics addressed include adaptive least mean square (LMS) filtering, adaptive filtering with discrete cosine transform LMS (DCT/LMS), adaptive noise cancelling, fetal electrocardiography, adaptive echo cancelling, inverse plant modeling, adaptive inverse con... View full abstract»

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