Showing papers in "Neural Networks in 1995"

TL;DR: In this article, a transiently chaotic neural network (TCNN) model is proposed for combinatorial optimization problems, where the chaotic neurodynamics is temporarily generated for searching and self-organizing, and eventually vanishes with autonomous decrease of a bifurcation parameter corresponding to the temperature in the usual annealing process.

TL;DR: A unified information geometrical framework for studying stochastic models of neural networks, by focusing on the EM and em algorithms, and proves a condition that guarantees their equivalence.

TL;DR: A neural network is proposed that recovers some original random signals from their linear mixtures observed by the same number of sensors and acquires the function with a learning process without using any particular information about the statistical properties of the sources and the coefficients of the linear transformation.

TL;DR: This framework study more closely generalizations of the problems of variance maximization and mean-square error minimization and derive gradient-type neural learning algorithms both for symmetric and hierarchic PCA-type networks.

TL;DR: Radiologists' reading procedure was modelled in order to instruct the artificial neural network to recognize the predefined image patterns and those of interest to experts and an unconventional method of using rotation and shift invariance is proposed to enhance the neural net performance.

TL;DR: It is shown that the EM algorithm can be regarded as a variable metric algorithm with its searching direction having a positive projection on the gradient of the log likelihood and an acceleration technique that yields a significant speedup in simulation experiments.

TL;DR: A model of a topologically organized neural network of a Hopfield type with nonlinear analog neurons is shown to be very effective for path planning and obstacle avoidance.

TL;DR: In the present study, genetic algorithms are proposed to automatically configure RBF networks and the network configuration is formed as a subset selection problem to find an optimal subset of nc terms from the Nt training data samples.

TL;DR: For the first time, different adaptive critic designs, a conventional proportional integral derivative (PID) regulator and backpropagation of utility are compared for the same control problem—automatic aircraft landing.

TL;DR: A model of factorial learning for general nonlinear transformations of an arbitrary non-Gaussian (or Gaussian) environment with statistically nonlinearly correlated input is presented.

TL;DR: A pseudo-objective function is formulated for the optimization problem in the form of a Lyapunov function to ensure the global convergence and the stability of the neural dynamic system by adopting an exterior penalty function method.

TL;DR: It is shown that SCS learning rates can be interpreted in terms of statistical decision theory, and several relationships between SCS and FLVQ are derived, which shows that the learning rates of these two algorithms have opposite tendencies.

TL;DR: A new algorithm for generating radial basis function (RBF)-like nets for classification problems using linear programming models to train the RBF-like net is presented.

TL;DR: A neural network model of boundary segmentation and surface representation is developed to process images containing range data gathered by a synthetic aperture radar (SAR) sensor and is shown to perform favorably in comparison to several other techniques for speckle removal.

TL;DR: Associative recall experiments on two pattern sets show that, besides the advantages of fast learning, guaranteed perfect storage, and full memory capacity, ARAM produces a stronger noise immunity than Bidirectional Associative Memory (BAM).

TL;DR: A hierarchy of macrodynamical equations for the recalling process of the autocorrelation associative memory model with synchronous dynamics is proposed and takes account of direct correlations between the cross talk noise terms at different time steps.

TL;DR: A learning rule of neural networks via a simultaneous perturbation and an analog feedforward neural network circuit using the learning rule, which requires only forward operations of the neural network and is suitable for hardware implementation.

TL;DR: Artificial neural networks have proven to be an interesting and useful alternate processing strategy for automatic target recognition (ATR) and the relation of neural classifiers to Bayesian techniques is emphasized along with the more recent use of feature sequences to enhance classification.

TL;DR: Radial basis function neural network architectures are introduced for the non linear adaptive noise cancellation problem and it is shown that by exploiting the duality with system identification, the nonlinear IIR filter can be configured as a recurrent radial basis function network.

TL;DR: Results show that properties that confer useful advantages for classification problems do not necessarily confer similar advantages for noisy mapping problems, and one particular feature, match tracking, is found to cause overlearning of the data.

TL;DR: The properties described in the paper are distinguished into a number of categories, as well as properties related to the number of list presentations needed for weight stabilization, which provide numerous insights as to how Fuzzy ART operates.

TL;DR: In summary, existence proofs derived from approximation theory prove to be irrelevant when the numerical limitations imposed by computer simulation are taken into account.

TL;DR: A proof that the settling dynamics for amean-field DUBM cause convergence to a free-energy minimum is presented, and a learning algorithm and simulations that demonstrate a mean-fieldDUBM's ability to learn interesting mappings are described.

TL;DR: Its performance, in terms of learning speed and scalability properties, is evaluated and found superior to the performance of reputedly fast variants of the back-propagation algorithm in the above benchmarks.

TL;DR: The present VIEWNET preprocessor includes the CORT-X 2 filter, which discounts the illuminant, regularizes and completes figural boundaries, and suppresses image noise, and this boundary segmentation is rendered invariant under 2-D translation, rotation, and dilation by use of a log polar transform.

TL;DR: A modular multi-stage architecture for focus-of-attention cueing, feature discovery and extraction, and one-class pattern learning and identification in synthetic aperture radar imagery is described.

TL;DR: A review of the four generic architectures for vision will be presented, providing a context for the term “ active vision”, and a justification for the importance, and the connection between, space-variant architectures and active vision methods.

TL;DR: The MLSOFM combines the ideas of self-organization and topographic mapping with those of multiscale image segmentation, and is formulated as one of vector quantization and is mapped onto the MLSSOFM.

TL;DR: A neural network was trained to recognize two three-dimensional shapes independent of orientation, based on echoes of ultrasonic pulses similar to those used by an echolocating bat, Eptesicus fuscus, and the network was required to generalize and recognize echoes from the shapes at novel orientations.

TL;DR: It is shown how models of contrast enhancement, contour, shading and color vision can be used to enhance targets in multispectral IR and SAR imagery, aiding in target detection.