Showing papers in "Neural Networks in 1999"

TL;DR: The bounds for convergence on learning-rate and momentum parameters are derived, and it is demonstrated that the momentum term can increase the range of learning rate over which the system converges.

TL;DR: Simulation results for both artificial and real data show remarkable improvement of generalization errors, supporting the idea of modifying a kernel function to enlarge the spatial resolution around the separating boundary surface by a conformal mapping, such that the separability between classes is increased.

TL;DR: This paper investigates how the learning modules specialized for these three kinds of learning can be assembled into goal-oriented behaving systems and presents a novel view that their computational roles can be characterized by asking what are the "goals" of their computation.

TL;DR: The experimental results show that negative correlation learning can produce neural network ensembles with good generalisation ability.

TL;DR: It is shown that if the space of mixing functions is not limited there exists always an infinity of solutions, and that for two dimensions, the solution is unique up to a rotation, if the mixing function is constrained to be a conformal mapping together with some other assumptions.

TL;DR: This article examines how model selection in neural networks can be guided by statistical procedures such as hypothesis tests, information criteria and cross validation, and proposes five specification strategies based on different statistical procedures.

TL;DR: An adaptive neuro-fuzzy system, HyFIS (Hybrid neural Fuzzy Inference System), for building and optimising fuzzy models and provides linguistic meaning to the connectionist architectures is proposed.

TL;DR: This paper formulates and studies a model of delayed impulsive Hopfield neural networks, which establishes several fundamental issues such as global exponential stability, existence and uniqueness of the equilibrium of such networks.

TL;DR: An on-line learning scheme is developed-the so-called mixture of recurrent neural net (RNN) experts-in which a set of RNN modules become self-organized as experts on multiple levels, in order to account for the different categories of sensory-motor flow which the robot experiences.

TL;DR: The main emphasis of the research presented here is on morphological bidirectional associative memories (MBAMs), which establish a mathematical theory for MBAMs and provide conditions that guarantee perfect biddirectional recall for corrupted patterns.

TL;DR: The aim of the paper is to investigate the application of control schemes based on "internal models" to the stabilization of the standing posture, showing that muscle stiffness alone is insufficient to carry out the task.

TL;DR: In this paper, Simon et al. used computational models to show how the face processing specialization apparently underlying prosopagnosia and visual object agnosia could be attributed to a relatively simple competitive selection mechanism that, during development, devotes neural resources to the tasks they are best at performing.

TL;DR: A universal acceleration technique for the BP algorithm based on extrapolation of each individual interconnection weight is presented, which suggests that it leads to significant savings in computation time of the standard BP algorithm.

TL;DR: Simulation results have shown that the use of the proposed learning algorithms avoids the instability problem and makes the ME architecture produce good performance in multiclass classification and the approximation algorithm leads to fast learning.

TL;DR: In this article, a model describing the activation dynamics of bidirectional associative memory (BAM) neural networks involving transmission delays was considered and a new notion, as a measure of restoring stability and termed it as a dead zone was introduced.

TL;DR: Inspired by CC, constructive backpropagation (CBP) is proposed and studied and it is shown that CBP is computationally just as efficient as the CC algorithm even though it needs to backpropagate the error through no more than one hidden layer.

TL;DR: Results on applying the evidence framework to the real-world data sets showed that committees of Bayesian networks achieved classification accuracies similar to the best alternative methods with a minimum of human intervention.

TL;DR: It is argued that a developmental approach does provide unique insights on how to build highly complex and adaptable artificial systems.

TL;DR: A new method of performing Canonical Correlation Analysis with Artificial Neural Networks is derived and is applied to Becker's random dot stereogram data and shown to be extremely effective at detecting shift information.

TL;DR: A new self-organizing neural network, the Kohonen Network Incorporating Explicit Statistics (KNIES) that is based on Kohonen's Self-Organizing Map (SOM), which has been used to solve the Euclidean Travelling Salesman Problem.

TL;DR: The measures are found to do an imperfect, but an adequate job in selecting a topographically optimal output space dimension, while they consistently single out particular maps as non-topographic.

TL;DR: The results show that the proposed system can significantly increase the accuracy of the product profile and is able to learn from the experience.

TL;DR: The nature of the basins of attraction of a Hopfield network is as important as the capacity, and a new learning rule is re-introduced, which has a higher capacity than that of the Hebb rule, and still keeps important functionality, such as incrementality and locality, which the pseudo-inverse lacks.

TL;DR: This work proposes a new architecture that maintains spatial relations between input features in winner-take-all WTA networks and analyzes the speed of selection, and shows that a two-stage selection network gains efficiency by combining selection with parallel removal of noisy regions.

TL;DR: In this paper, the anatomically defined concept of columns is placed into a functional context by a large-scale computational hypothesis on visual recognition, which includes both, rapid parallel forward recognition and a feedback controlled refinement system.

TL;DR: This paper considers the correlation coefficient of consecutive ISIs, which was ignored in previous studies, and concludes that the leaky integrate-and-fire model with temporally correlated inputs does account for the biological data.

TL;DR: The improved BAM model opens interesting new perspectives in information retrieval it allows efficient data access providing segmentation of ambiguous user input, relevance feedback and relevance ranking, and the implication of this for a more flexible version of Hebbian cell-assemblies.

TL;DR: The article presents some ideas regarding weighting of multiple agents and extends them into partitioning an input/state space into multiple regions with differential weighting to reduce the learning complexity of agents (and their function approximators) and thus to facilitate the learning overall.

TL;DR: A neural network system for boundary segmentation and surface representation, inspired by a new local-circuit model of visual processing in the cerebral cortex, is used to enhance images of range data gathered by a synthetic aperture radar (SAR) sensor.

TL;DR: Under certain conditions, it is shown that the mixtures-of-experts systems are identifiable if the experts are ordered and the gating parameters are initialized.