Showing papers in "Neurocomputing in 1999"

TL;DR: SpikeNET is a simulator for modeling large networks of asynchronously spiking neurons which undergo step-like changes in membrane potential when synaptic inputs arrive and is used to model complex multi-layer architectures based on the primate visual system that involve millions of neurons and billions of synaptic connections.

TL;DR: It is proved that there exists an analytic, strictly monotone, sigmoidal activation function for which this lower bound is essentially attained and that one can approximate arbitrarily well any continuous function on any compact domain by a two hidden layer MLP using a fixed finite number of units in each layer.

TL;DR: In this study neural network and genetic algorithm fuzzy rule induction systems have been developed and applied to three classification problems and it is indicated that the genetic/fuzzy approach compares more than favourably with the neuro/ fuzzy and rough set approaches.

TL;DR: It is shown that models built from bootstrap aggregated neural networks are more accurate and robust than those built from single neural networks.

TL;DR: Numerical solutions of the coupled non-linear system of partial differential equations show properties analogous to cortical evoked potentials, oscillations at the frequency of the mammalian alpha rhythm and non-stationary epileptic spikes.

TL;DR: Various neural network architectures and associated adaptive learning algorithms are discussed for handling the cases where the number of sources is unknown, and techniques include estimation of thenumber of sources, redundancy removal among the outputs of the networks, and extraction of the sources one at a time.

TL;DR: In this paper, a fast algorithm is presented for the training of multilayer perceptron neural networks, which uses separate error functions for each hidden unit and solves multiple sets of linear equations.

TL;DR: The author describes a novel behavior-based, fully distributed architecture, called ALLIANCE, that utilizes adaptive action selection to achieve fault tolerant cooperative control in robot missions involving loosely coupled, largely independent tasks.

TL;DR: It is demonstrated that techniques for tracking the recorded neural population over time through analysis of unit waveforms, principle component clusters and response properties of single units and multiunit clusters can be used to provide a useful measure of unit stability over extended recording periods.

TL;DR: Based on the neural network model, a controller with extended control horizon is developed and the implementation issues are discussed, with particular emphasis on an efficient quasi-Newton algorithm.

TL;DR: Simultaneous single or paired intracellular and tetrode recordings are used here to evaluate a contemporary spike sorting algorithm for isolated as well as overlapped events and to demonstrate that overlapping extracellular spikes combine additively.

TL;DR: A novel heuristics approach for neural networks global learning algorithm based upon the least-squares method and a Penalty approach to solve the problem of local minima that outperforms other conventional algorithms in terms of convergence speed and the ability of escaping from theLocal minima.

TL;DR: This work considers a rule-based fuzzy controller and a learning procedure based on the stochastic approximation method and shows that a modified form of this network is identical with the fuzzy controller, which may thus be considered as a neuro-fuzzy controller.

TL;DR: An experimental study of the certainty neuron fuzzy cognitive maps (CNFCMs) dynamical behaviour is presented as this appears through simulations as the system uses a two variable transfer function that provides them with memory capabilities and decay mechanism.

TL;DR: Parsimonious DRNN models are able to find an appropriate internal representation of various chaotic processes from the observation of a subset of the state variables of a dynamical system.

TL;DR: Three related measures of input parameter influence can be used to support explanation facilities for neural networks and an algorithm for generating rules from real-valued networks based on the influence measures is presented.

TL;DR: A system formed by a mixture of expert models (MEM) for time-series forecasting with improvements in forecast performance when compared with the single models as experimentally demonstrated through two different time series: laser data and exchange rate data.

TL;DR: The results show that the original RTRL algorithm achieves the lowest error among the gradient-based algorithms, but it requires the longest training time; whereas the sub-grouping strategy uses the shortest training time but its convergence capability is the poorest.

TL;DR: In a Banach space X satisfying mild conditions, for its infinite, linearly independent subset G, there is no continuous best approximation map from X to the n-span, span n G.

TL;DR: Psychophysics experiments on the human visual system have established that the sensitivity for the detection of fine detail in noise contaminated images can be quantitatively and repeatably measured using stochastic resonance as a tool, but what does this mean?

TL;DR: The concept of a dynamic synapse is extended to provide a formalism to incorporate synaptic mechanisms into a general scheme of neural information processing and can be concisely expressed and interpreted in these two synaptic terms.

TL;DR: This framework for discussing the generalization ability of a trained network in the original function space using tools of functional analysis based on reproducing kernel Hilbert spaces is introduced and provides a useful hint in performing kernel-based approximations in an online setting.

TL;DR: The INSS system is presented, a new hybrid approach based upon the principles of KBANN networks that offers a new approach applicable to constructive machine learning with high-performance tools, even in the presence of incomplete or erroneous data.

TL;DR: An array processing technique is developed to remove the correlated component of noise, improving the signal-to-noise ratio and it is shown that the weighting vector of the array algorithm can be manipulated to facilitate sorting of the action potentials.

TL;DR: With the presented tuning schema, interaction of RE and dialog module becomes possible even before tests with blind subjects, and tunable spatiotemporal functions and corresponding perception-based dialog procedure for a retina implant as visual prosthesis are proposed.

TL;DR: A new algorithm for learning invariance manifolds is introduced that allows a neuron to learn a non-linear input–output function to extract invariant or rather slowly varying features from a vectorial input sequence.

TL;DR: This paper provides an approach for output feedback robust approximate pole assignment formulated as an unconstrained optimization problem and solved via the gradient flow approach which is ideally suited for neural computing implementation.

TL;DR: This work proposes the use of a hyperellipsoidal thresholding surface in the four-dimensional space of the signal values to detect spikes, which provides a better approximation of the equiprobable surface of the noise amplitude distribution compared to the traditionally used hypercubical thresholded surface.

TL;DR: This work develops a set of biologically constrained tools for constructing networks which exhibit interesting behavior and provides an example of the application of these tools to modeling the part of the brain which controls horizontal eye position.

TL;DR: The model has enabled us to explore features of the coincidence detector neurons unexplained by a simpler biophysical model, including the effect of synapse location and multiple dendrites.