By Topic

[Proceedings 1992] IJCNN International Joint Conference on Neural Networks

7-11 Jun 1992

Go

Filter Results

Displaying Results 1 - 25 of 138
  • Learning probabilities for causal networks

    Publication Year: 1992, Page(s):97 - 102 vol.4
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (448 KB)

    The author presents an unsupervised method to learn probabilities of random events. Learning is done by letting variables adaptively respond to positive and negative environmental stimuli. The basic learning rule is applied to learn prior and conditional probabilities for causal networks. By combining with a stochastic factor, this method is extended to learn probabilities of hidden causations, a ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The N-N-N conjecture in ART1

    Publication Year: 1992, Page(s):103 - 108 vol.4
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (384 KB)

    The authors consider the ART1 neural network architecture introduced by G.A. Carpenter and S. Grossberg (Comput. Vis., Graph., and Image Process. vol.37, 54-115, 1987). In their original paper, Carpenter and Grossberg made the following conjecture. In the fast learning case if the F2 layer in ART1 has at least N nodes, then each member of a list of N input pat... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A neural network model of spatio-temporal pattern recognition, recall, and timing

    Publication Year: 1992, Page(s):109 - 114 vol.4
    Cited by:  Papers (1)  |  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (480 KB)

    The author describes the design of a self-organizing, hierarchical network model of unsupervised serial learning. The model learns to recognize, store, and recall sequences of unitized patterns, using either short-term memory (STM) or both STM and long-term memory (LTM) mechanisms. Timing information is learned and recall both from STM and from LTM is performed with a learned rhythmical structure.... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Neurodynamical model of collective brain

    Publication Year: 1992, Page(s):115 - 121 vol.4
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (388 KB)

    A dynamical system which mimics collective purposeful activities of a set of units of intelligence is introduced and discussed. A global control of the unit activities is replaced by the probabilistic correlations between them. These correlations are learned during a long term period of performing collective tasks, and are stored in the synaptic interconnections. The model is represented by a syst... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Resolving the components of transient signals using neural network and subspace inhibition filter algorithms

    Publication Year: 1992, Page(s):283 - 288 vol.4
    Cited by:  Papers (3)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (324 KB)

    The authors compare the use of a multilayer perceptron neural network with the use of a subspace inhibition filter algorithm for achieving enhanced local resolution of the arrival times and frequencies of the components of a transient signal. Both the neural network algorithm and the subspace inhibition filter algorithm provide enhanced local resolution of the arrival times and frequencies of the ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An optimized filter architecture incorporating a neural net

    Publication Year: 1992, Page(s):543 - 548 vol.4
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (496 KB)

    The Kalman filtering process is generally a combination of predicting the state variables and then correcting them with the measurement data from the sensor systems. The approach presented here is to use a neural network to make the corrections to the state variable estimates within a Kalman filter structure. A model is used for the prediction of the state variable estimates and corrections are ma... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Learning of continuously transformed pattern cycles by an oscillatory neural network

    Publication Year: 1992, Page(s):122 - 127 vol.4
    Cited by:  Papers (4)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (408 KB)

    A cognitive psychological pattern transformation in the processes of the brain is discussed. It is assumed that the transformation is processed by physiological oscillations. An oscillatory neural network is proposed to embed the continuously transformed pattern cycles in the limit cycles. The oscillation by destabilized memory points and the characteristic of a dynamic trajectory for an input pat... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Dim target detection and clutter rejection using modified high order correlation neural network

    Publication Year: 1992, Page(s):289 - 294 vol.4
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (396 KB)

    The authors present a scheme for detecting dim moving targets in highly cluttered background from infrared (IR) data. A high-order spatiotemporal correlation scheme developed by R.J. Liou et al. (1991) to extract the sequency information carried by a target track and reject the background clutter is modified to incorporate target motion dynamics. More than 97% clutter rejection is achieved without... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Regression analysis of spectroscopic process data using a combined architecture of linear and nonlinear artificial neural networks

    Publication Year: 1992, Page(s):549 - 554 vol.4
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (476 KB)

    The authors demonstrate that a combined architecture of linear and nonlinear artificial neural networks offers many advantages over the conventional multilayer feedforward networks and the conventional biased regression methods for the modeling of spectroscopic process data. This direct linear feedthrough (DLF) network is an especially useful tool for modeling process data when the true linear or ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Discriminative training algorithm for predictive neural network models

    Publication Year: 1992, Page(s):685 - 690 vol.4
    Cited by:  Papers (5)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (272 KB)

    A discriminative training algorithm for predictive neural network models is proposed. The algorithm is applied to a speaker independent isolated digit recognition experiment. The recognition error rate is reduced from 2.52% when the classifier is trained with a non-discriminative algorithm to 0.58% when the discriminative algorithm is applied. The increase in classifier discrimination ability is a... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Correlation matrices and the construction of successful network recodings

    Publication Year: 1992, Page(s):808 - 813 vol.4
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (344 KB)

    Previous work performed by the authors demonstrated the importance of the amount of synaptic connectivity and output firing thresholds, and the relative unimportance of an activity based weight modification rule, in the construction of neural networks which perform successful recodings. The authors attempt to construct near optimal recoding networks incorporating and building upon the results of p... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Handwritten alphabet and digit character recognition using feature extracting neural network and modified self-organizing map

    Publication Year: 1992, Page(s):235 - 240 vol.4
    Cited by:  Papers (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (412 KB)

    A new pattern recognition method is proposed for handwritten alphabet and digit characteristics. The feature point distribution of a standard pattern is mapped onto that of a distorted pattern, through a modified self-organizing map (SOM). The distorted pattern is recognized based on similarity between both feature point distributions. The modified SOM has the following advantages. First, the numb... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Estimation of oscillation frequency in coupled cell assemblies

    Publication Year: 1992, Page(s):128 - 133 vol.4
    Cited by:  Papers (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (348 KB)

    A macroscopic model of cell assemblies is used to analyze the ensemble behavior of coupled sub-populations of inhibitory and excitatory neurons. For such systems, rhythmic behavior can occur even in the absence of weight change dynamics. Conditions that give rise to such oscillatory behavior through the interaction of two neural groups are identified. Using Fourier series analysis, the authors obt... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Error correcting neural networks for channels with Gaussian noise

    Publication Year: 1992, Page(s):295 - 300 vol.4
    Cited by:  Patents (4)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (356 KB)

    The main features of error correcting codes and standard decoding techniques are reviewed. Feedforward neural networks for soft-decision decoding of block codes in channels with additive white Gaussian noise are presented. When the noise is not white, the authors deduce the optimal set of weights for the connections of the network. These weights are also approximately obtained by an error backprop... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Data-parallel training of spatiotemporal connectionist networks on the Connection Machine

    Publication Year: 1992, Page(s):555 - 559 vol.4
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (276 KB)

    An algorithm for optimizing spatiotemporal connectionist networks utilizing training set parallelism has been implemented on the Connection Machine (CM). The algorithm supports several optimization methods including backpropagation, conjugate gradient, and pseudo-Newtonian. By allocating one CM processor per training example, the computational complexity of the gradient derivation becomes independ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Weight-space probability densities and convergence times for stochastic learning

    Publication Year: 1992, Page(s):158 - 164 vol.4
    Cited by:  Papers (4)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (452 KB)

    The authors extend the theory of search dynamics for stochastic learning algorithms, address the time evolution of the weight-space probability density and the distribution of convergence times, with particular attention given to escape from local optima, and develop a theoretical framework that describes the evolution of the weight-space probability density. The primary results are exact predicti... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A neural network approach to large dimensional spectral pattern processing

    Publication Year: 1992, Page(s):691 - 696 vol.4
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (512 KB)

    The authors present a multiple neural network system that extracts and interprets spatiotemporal features from two-dimensional spectral images. The system uses interconnected multiple networks where the first network extracts spatial features and successive networks label and classify the features. The labeling network uses a priori knowledge on its connection weights, thereby eliminating the need... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Genetic breeding of control parameters for the Hopfield/Tank neural net

    Publication Year: 1992, Page(s):618 - 623 vol.4
    Cited by:  Papers (4)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (356 KB)

    The authors show how genetic algorithms may be used in conjunction with the Hopfield/Tank neural net by breeding an effective set of control parameters in the parameter sub-space to be used by the artificial neural network. They briefly consider the standard Hopfield/Tank neural net followed by a discussion of the genetic algorithm used with this network. Some of the more important operators used ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Dystal: a self-organizing ANN with pattern independent training time

    Publication Year: 1992, Page(s):814 - 819 vol.4
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (380 KB)

    As the difficulty of problems increase, artificial neural networks (ANNs) that use nonlinear optimization suffer from degraded execution speed, particularly with respect to learning time. Dystal is an ANN which does not suffer this degradation. Dystal is an ANN based on properties of associative learning found in biological neural networks. To verify these theoretical properties of Dystal, the aut... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Training recurrent networks using the extended Kalman filter

    Publication Year: 1992, Page(s):241 - 246 vol.4
    Cited by:  Papers (43)  |  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (452 KB)

    The author describes some relationships between the extended Kalman filter (EKF) as applied to recurrent net learning and some simpler techniques that are more widely used. In particular, making certain simplifications to the EKF gives rise to an algorithm essentially identical to the real-time recurrent learning (RTRL) algorithm. Since the EKF involves adjusting unit activity in the network, it a... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Recurrent competitive Hebbian learning

    Publication Year: 1992, Page(s):767 - 772 vol.4
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (468 KB)

    Competitive Hebbian learning is extended to networks with trainable lateral connections, in addition to the trainable feedforward connections considered previously by the author (1991,1992). These recurrent systems are able to learn to respond to ordering in time of the input vectors. The theoretical framework for the extension of competitive Hebbian learning to recurrent systems is presented. Thi... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Sentence processing with realistic feedback

    Publication Year: 1992, Page(s):661 - 666 vol.4
    Cited by:  Papers (6)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (408 KB)

    The author presents a connectionist natural language processing model, SAIL1, which uses a recurrent network topology to process English sentences. SAIL1 will build the sentence meaning representation incrementally, incorporating into the meaning only the information derived from words prior to the current word. The network is trained only on that part of the sentence meaning representing the data... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An efficient neural network architecture for recognition of spatial pattern invariants

    Publication Year: 1992, Page(s):208 - 213 vol.4
    Cited by:  Papers (4)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (364 KB)

    The authors describe a neural architecture for efficient recognition of invariant features placed arbitrarily in patterns of data. The architecture provides versatility in invariant selection with minimal computation and storage requirements. Operating in dumb mode, the architecture, called the big, dumb mass detector (BDMD), autonomously extracts fixed-size subsets of pattern components based upo... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Fixed point analysis for discrete-time recurrent neural networks

    Publication Year: 1992, Page(s):134 - 139 vol.4
    Cited by:  Papers (13)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (332 KB)

    The author shows the existence of a fixed point for every recurrent neural network and uses a geometric approach to locate where the fixed points are. The stability is discussed for low-gain and high-gain situations. A generalized Hopfield saturation theorem is presented in a high gain situation for a discrete-time model version View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Artificial neural networks for 3D nonrigid motion analysis

    Publication Year: 1992, Page(s):420 - 425 vol.4
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (524 KB)

    A novel approach to 3D nonrigid motion analysis using artificial neural networks is presented. A set of neural networks is proposed to tackle the problem of nonrigidity in 3D motion estimation. Constraints are specified to ensure a stable and global consistent estimation of local deformations. The assignments of weights between two layers, the initial values of the outputs, and the connections bet... View full abstract»

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