SEARCH · PubMed Health

Results for “radial basis function network”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 271 records · Page 15Linked to original sources

PubMed1998

The diabolo classifier

We present a new classification architecture based on autoassociative neural networks that are used to learn discriminant models of each class. The proposed architecture has several interesting properties with respect to other model-based classifiers like nearest-neighbors or radial basis functions: it has a low computational complexity and uses a compact distributed representation of the models. The classifier is also well suited for the incorporation of a priori knowledge by means of a problem-specific distance measure. In particular, we will show that tangent distance (Simard, Le Cun, & Denker, 1993) can be used to achieve transformation invariance during learning and recognition. We demonstrate the application of this classifier to optical character recognition, where it has achieved state-of-the-art results on several reference databases. Relations to other models, in particular those based on principal component analysis, are also discussed.

Journal Article↗

PubMed1997

Symmetry breaking and training from incomplete data with Radial Basis Boltzmann Machines.

A Radial Basis Boltzmann Machine (RBBM) is a specialized Boltzmann Machine architecture that combines feed-forward mapping with probability estimation in the input space, and for which very efficient learning rules exist. The hidden representation of the network displays symmetry breaking as a function of the noise in the dynamics. Thus, generalization can be studied as a function of the noise in the neuron dynamics instead of as a function of the number of hidden units. We show that the RBBM can be seen as an elegant alternative of k-nearest neighbor, leading to comparable performance without the need to store all data. We show that the RBBM has good classification performance compared to the MLP. The main advantage of the RBBM is that simultaneously with the input-output mapping, a model of the input space is obtained which can be used for learning with missing values. We derive learning rules for the case of incomplete data, and show that they perform better on incomplete data than the traditional learning rules on a 'repaired' data set.

Computer Simulation↗

PubMed2006

Locally weighted interpolating growing neural gas.

In this paper, we propose a new approach to function approximation based on a growing neural gas (GNG), a self-organizing map (SOM) which is able to adapt to the local dimension of a possible high-dimensional input distribution. Local models are built interpolating between values associated with the map's neurons. These models are combined using a weighted sum to yield the final approximation value. The values, the positions, and the "local ranges" of the neurons are adapted to improve the approximation quality. The method is able to adapt to changing target functions and to follow nonstationary input distributions. The new approach is compared to the radial basis function (RBF) extension of the growing neural gas and to locally weighted projection regression (LWPR), a state-of-the-art algorithm for incremental nonlinear function approximation.

Algorithms↗

PubMed2004

Asymmetric kernel regression.

Kernel regression is one model that has been applied to explain or design radial-basis neural networks. Practical application of the kernel regression method has shown that bias errors caused by the boundaries of the data can seriously effect the accuracy of this type of regression. This paper investigates the correction of boundary error by substituting an asymmetric kernel function for the symmetric kernel function at data points close to the boundary. The asymmetric kernel function allows a much closer approach to the boundary to be achieved without adversely effecting the noise-filtering properties of the kernel regression.

Neural Networks, Computer↗

PubMed1987

Astrocytes in the postnatal central nervous system. From birth to 14 years of age. An immunohistochemical study on paraffin-embedded material.

The postnatal development of the fibrillary astrocyte in 25 children without any anamnestic or post mortem signs of disease were studied. The investigation was performed on basis of the immunoperoxidase method for glial fibrillary acidic protein (GFAP). The supporting function of the external glial limiting membrane, the glial vascular limiting membrane and the metabolic role of the astrocytes are stressed. During the postnatal myelination of the white matter the immature astrocyte changed into the mature form and a dense fibrous network developed. The radially oriented glial processes of the Bergmann astrocytes of the cerebellum guided the fetal external granular cells during the migration. In the brainstem we found minor variations of the dense fibrous astrocytic network both in children dead after accidents and in sudden infant death syndrome. These findings presumably demonstrate a normal pattern.

Adolescent↗

PubMed1998

Spatial and temporal pattern analysis via spiking neurons.

Spiking neurons, receiving temporally encoded inputs, can compute radial basis functions (RBFs) by storing the relevant information in their delays. In this paper we show how these delays can be learned using exclusively locally available information (basically the time difference between the pre- and postsynaptic spikes). Our approach gives rise to a biologically plausible algorithm for finding clusters in a high-dimensional input space with networks of spiking neurons, even if the environment is changing dynamically. Furthermore, we show that our learning mechanism makes it possible that such RBF neurons can perform some kind of feature extraction where they recognize that only certain input coordinates carry relevant information. Finally we demonstrate that this model allows the recognition of temporal sequences even if they are distorted in various ways.

Action Potentials↗

PubMed2006

Multiuser receiver for DS-CDMA signals in multipath channels: an enhanced multisurface method.

This paper deals with the problem of multiuser detection in direct-sequence code-division multiple-access (DS-CDMA) systems in multipath environments. The existing multiuser detectors can be divided into two categories: (1) low-complexity poor-performance linear detectors and (2) high-complexity good-performance nonlinear detectors. In particular, in channels where the orthogonality of the code sequences is destroyed by multipath, detectors with linear complexity perform much worse than the nonlinear detectors. In this paper, we propose an enhanced multisurface method (EMSM) for multiuser detection in multipath channels. EMSM is an intermediate piecewise linear detection scheme with a run-time complexity linear in the number of users. Its bit error rate performance is compared with existing linear detectors, a nonlinear radial basis function detector trained by the new support vector learning algorithm, and Verdu's optimal detector. Simulations in multipath channels, for both synchronous and asynchronous cases, indicate that it always outperforms all other linear detectors, performing nearly as well as nonlinear detectors.

Algorithms↗

PubMed2005

Support vector machines for automated gait classification.

Ageing influences gait patterns causing constant threats to control of locomotor balance. Automated recognition of gait changes has many advantages including, early identification of at-risk gait and monitoring the progress of treatment outcomes. In this paper, we apply an artificial intelligence technique [support vector machines (SVM)] for the automatic recognition of young-old gait types from their respective gait-patterns. Minimum foot clearance (MFC) data of 30 young and 28 elderly participants were analyzed using a PEAK-2D motion analysis system during a 20-min continuous walk on a treadmill at self-selected walking speed. Gait features extracted from individual MFC histogram-plot and Poincaré-plot images were used to train the SVM. Cross-validation test results indicate that the generalization performance of the SVM was on average 83.3% (+/-2.9) to recognize young and elderly gait patterns, compared to a neural network's accuracy of 75.0+/-5.0%. A "hill-climbing" feature selection algorithm demonstrated that a small subset (3-5) of gait features extracted from MFC plots could differentiate the gait patterns with 90% accuracy. Performance of the gait classifier was evaluated using areas under the receiver operating characteristic plots. Improved performance of the classifier was evident when trained with reduced number of selected good features and with radial basis function kernel. These results suggest that SVMs can function as an efficient gait classifier for recognition of young and elderly gait patterns, and has the potential for wider applications in gait identification for falls-risk minimization in the elderly.

Adult↗

PubMed1995

Pattern recognition computation using action potential timing for stimulus representation.

A computational model is described in which the sizes of variables are represented by the explicit times at which action potentials occur, rather than by the more usual 'firing rate' of neurons. The comparison of patterns over sets of analogue variables is done by a network using different delays for different information paths. This mode of computation explains how one scheme of neuroarchitecture can be used for very different sensory modalities and seemingly different computations. The oscillations and anatomy of the mammalian olfactory systems have a simple interpretation in terms of this representation, and relate to processing in the auditory system. Single-electrode recording would not detect such neural computing. Recognition 'units' in this style respond more like radial basis function units than elementary sigmoid units.

Action Potentials↗

PubMed1998

The connection between regularization operators and support vector kernels.

In this paper a correspondence is derived between regularization operators used in regularization networks and support vector kernels. We prove that the Green's Functions associated with regularization operators are suitable support vector kernels with equivalent regularization properties. Moreover, the paper provides an analysis of currently used support vector kernels in the view of regularization theory and corresponding operators associated with the classes of both polynomial kernels and translation invariant kernels. The latter are also analyzed on periodical domains. As a by-product we show that a large number of radial basis functions, namely conditionally positive definite functions, may be used as support vector kernels.

Journal Article↗

PubMed2003

[Quick identification of scullcaps in different geographical origins using clustering analysis method and infrared fingerprint spectra].

Based on the infrared fingerprint spectra, scullcap samples from 15 geographical origins were clustered into 6 classes with principal component analysis. The classification was related to their geographical origins and weather. In the same class, chemical components of scullcaps are similar to each other, which can be considered as the criterion of evaluating the quality of scullcaps. 43 geographical origins were predicted with radial basis function, which was demonstrated to be a powerful ANN method in discrimination. The classes of scullcaps can be distinguished by the method. The method can be used to class and identify geographical origins of scullcaps. It is one of the methods in the quality control of traditional Chinese medicines.

Cluster Analysis↗

PubMed1995

Learning to ignore: psychophysics and computational modeling of fast learning of direction in noisy motion stimuli.

The effects of practice on the discrimination of direction of motion in briefly presented noisy dynamic random dot patterns are investigated in several forced-choice psychophysical tasks. We found that the percentage of correct responses on any specific task increases linearly with repetition of trials within roughly 200 trials from about chance to a performance of 90% or better. The level of performance remained constant or improved over several days, and in most instances it did not transfer when stimulus parameters changed. We used a modified Radial Basis Function (RBF) representation to model the psychophysical tasks. The performance of the model is functionally similar to the psychophysical results. We propose a Hebbian learning algorithm which deactivates the inputs from neurons responding to motion noise in the stimulus. Our computational model suggests that to solve this task in biological systems, neurons (perhaps in MT) improve their performance by 'learning to ignore' noise in the image.

Acoustic Stimulation↗

PubMed2001

Robust full Bayesian learning for radial basis networks.

We propose a hierarchical full Bayesian model for radial basis networks. This model treats the model dimension (number of neurons), model parameters, regularization parameters, and noise parameters as unknown random variables. We develop a reversible-jump Markov chain Monte Carlo (MCMC) method to perform the Bayesian computation. We find that the results obtained using this method are not only better than the ones reported previously, but also appear to be robust with respect to the prior specification. In addition, we propose a novel and computationally efficient reversible-jump MCMC simulated annealing algorithm to optimize neural networks. This algorithm enables us to maximize the joint posterior distribution of the network parameters and the number of basis function. It performs a global search in the joint space of the parameters and number of parameters, thereby surmounting the problem of local minima to a large extent. We show that by calibrating the full hierarchical Bayesian prior, we can obtain the classical Akaike information criterion, Bayesian information criterion, and minimum description length model selection criteria within a penalized likelihood framework. Finally, we present a geometric convergence theorem for the algorithm with homogeneous transition kernel and a convergence theorem for the reversible-jump MCMC simulated annealing method.

Journal Article↗

PubMedDate not supplied

BYY harmony learning, structural RPCL, and topological self-organizing on mixture models.

The Bayesian Ying-Yang (BYY) harmony learning acts as a general statistical learning framework, featured by not only new regularization techniques for parameter learning but also a new mechanism that implements model selection either automatically during parameter learning or via a new class of model selection criteria used after parameter learning. In this paper, further advances on BYY harmony learning by considering modular inner representations are presented in three parts. One consists of results on unsupervisedmixture models, ranging from Gaussian mixture based Mean Square Error (MSE) clustering, elliptic clustering, subspace clustering to NonGaussian mixture based clustering not only with each cluster represented via either Bernoulli-Gaussian mixtures or independent real factor models, but also with independent component analysis implicitly made on each cluster. The second consists of results on supervised mixture-of-experts (ME) models, including Gaussian ME, Radial Basis Function nets, and Kernel regressions. The third consists of two strategies for extending the above structural mixtures into self-organized topological maps. All these advances are introduced with details on three issues, namely, (a) adaptive learning algorithms, especially elliptic, subspace, and structural rival penalized competitive learning algorithms, with model selection made automatically during learning; (b) model selection criteria for being used after parameter learning, and (c) how these learning algorithms and criteria are obtained from typical special cases of BYY harmony learning.

Bayes Theorem↗

PubMed2006

Molecular dynamics simulations of trehalose as a 'dynamic reducer' for solvent water molecules in the hydration shell.

Systematic computational work for a series of 13 disaccharides was performed to provide an atomic-level insight of unique biochemical role of the alpha,alpha-(1-->1)-linked glucopyranoside dimer over the other glycosidically linked sugars. Superior osmotic and cryoprotective abilities of trehalose were explained on the basis of conformational and hydration characteristics of the trehalose molecule. Analyses of the hydration number and radial distribution function of solvent water molecules showed that there was very little hydration adjacent to the glycosidic oxygen of trehalose and that the dynamic conformation of trehalose was less flexible than any of the other sugars due to this anisotropic hydration. The remarkable conformational rigidity that allowed trehalose to act as a sugar template was required for stable interactions with hydrogen-bonded water molecules. Trehalose made an average of 2.8 long-lived hydrogen bonds per each MD step, which was much larger than the average of 2.1 for the other sugars. The stable hydrogen-bond network is derived from the formation of long-lived water bridges at the expense of decreasing the dynamics of the water molecules. Evidence for this dynamic reduction of water by trehalose was also established based on each of the lowest translational diffusion coefficients and the lowest intermolecular coulombic energy of the water molecules around trehalose. Overall results indicate that trehalose functions as a 'dynamic reducer' for solvent water molecules based on its anisotropic hydration and conformational rigidity, suggesting that macroscopic solvent properties could be modulated by changes in the type of glycosidic linkages in sugar molecules.

Computer Simulation↗

PubMed2001

Best harmony, unified RPCL and automated model selection for unsupervised and supervised learning on Gaussian mixtures, three-layer nets and ME-RBF-SVM models.

After introducing the fundamentals of BYY system and harmony learning, which has been developed in past several years as a unified statistical framework for parameter learning, regularization and model selection, we systematically discuss this BYY harmony learning on systems with discrete inner-representations. First, we shown that one special case leads to unsupervised learning on Gaussian mixture. We show how harmony learning not only leads us to the EM algorithm for maximum likelihood (ML) learning and the corresponding extended KMEAN algorithms for Mahalanobis clustering with criteria for selecting the number of Gaussians or clusters, but also provides us two new regularization techniques and a unified scheme that includes the previous rival penalized competitive learning (RPCL) as well as its various variants and extensions that performs model selection automatically during parameter learning. Moreover, as a by-product, we also get a new approach for determining a set of 'supporting vectors' for Parzen window density estimation. Second, we shown that other special cases lead to three typical supervised learning models with several new results. On three layer net, we get (i) a new regularized ML learning, (ii) a new criterion for selecting the number of hidden units, and (iii) a family of EM-like algorithms that combines harmony learning with new techniques of regularization. On the original and alternative models of mixture-of-expert (ME) as well as radial basis function (RBF) nets, we get not only a new type of criteria for selecting the number of experts or basis functions but also a new type of the EM-like algorithms that combines regularization techniques and RPCL learning for parameter learning with either least complexity nature on the original ME model or automated model selection on the alternative ME model and RBF nets. Moreover, all the results for the alternative ME model are also applied to other two popular nonparametric statistical approaches, namely kernel regression and supporting vector machine. Particularly, not only we get an easily implemented approach for determining the smoothing parameter in kernel regression, but also we get an alternative approach for deciding the set of supporting vectors in supporting vector machine.

Algorithms↗

PubMed1999

Classification of tetraplegics through automatic movement evaluation.

The general problem of classification of functional movements in humans with spinal cord injuries requires the following questions to be answered: what are the essential kinematic parameters that we have to observe during the movement? Is it possible to estimate preserved motor skills based on kinematics? Which computational method for identification is suited to geometric feature analysis? To answer these questions we have developed the methodology which has two phases: (1) recordings of a series of specified arm movements; and (2) custom made software for graphical presentation of arm movements and the design of wavelet and neural networks for movement classification. The proposed protocol is automated and both graphical presentation and neural networks allow easy interpretation of the instrumented assessment to accomplish automatic classification of arm movements in tetraplegics. The protocol was evaluated on 16 spinal cord injury (SCI) patients and seven healthy control subjects for three different arm movements. The classification rate yielded results in the range 46-100% for movement trials that were tested. The application of neural networks for classification of arm movements is completed with results using different neural networks: backpropagation, radial basis, recurrent (Elman), self-organizing and Learning Vector Quantization (LVQ).

Adolescent↗

PubMed2000

Structural evidence for ion transport and tectorial membrane maintenance in the gerbil limbus.

Cells medial to the tunnel of Corti were examined to assess fine structural features relevant to their proposed role in cochlear K(+) homeostasis. A dense network of canaliculi referred to as canalicular reticulum (CR) resided in the foot body of inner pillar cells, where it bordered and could resorb ions released from inner radial and spiral nerves. Lateral interdental cells (IDCs) formed columns which connected the inner sulcus epithelium with the base of the tectorial membrane's (TM) middle zone. A spout-like neck in cells at the top of lateral IDC columns housed a dense concentration of CR which resembled that characteristic of ion transporting epithelia and appeared to be located here for transporting ions and fluid toward the TM. Clustered IDCs in the center of the limbus connected underlying limbal stroma with the TM's limbal zone and appeared capable of transporting ions from stroma to TM. Abundant CR in limbal stellate fibrocytes evidenced their capacity to transport ions and fluid, presumably from inner sulcus epithelium toward central IDCs. The most medial IDCs possibly function as the terminus of an ion cycling path from scala vestibuli to endolymph. Light fibrocytes situated between supralimbal fibrocytes and medial IDCs appeared to serve as a link in this pathway. The limbal zone of the TM overlying central IDCs consisted of three distinct regions which offered a structural basis for transformation of an amorphous matrix supplied by central IDCs into the protofibrils of the membrane's middle zone.

Animals↗