Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Neural network”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

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

At least 631 records · Page 35Linked to original sources
PubMed2005

Neural-network-Based adaptive hybrid-reflectance model for 3-D surface reconstruction.

This paper proposes a novel neural-network-based adaptive hybrid-reflectance three-dimensional (3-D) surface reconstruction model. The neural network automatically combines the diffuse and specular components into a hybrid model. The proposed model considers the characteristics of each point and the variant albedo to prevent the reconstructed surface from being distorted. The neural network inputs are the pixel values of the two-dimensional images to be reconstructed. The normal vectors of the surface can then be obtained from the output of the neural network after supervised learning, where the illuminant direction does not have to be known in advance. Finally, the obtained normal vectors are applied to enforce integrability when reconstructing 3-D objects. Facial images and images of other general objects were used to test the proposed approach. The experimental results demonstrate that the proposed neural-network-based adaptive hybrid-reflectance model can be successfully applied to objects generally, and perform 3-D surface reconstruction better than some existing approaches.

Algorithms↗
PubMedDate not supplied

A regenerating spiking neural network.

Due to their distributed architecture, artificial neural networks often show a graceful performance degradation to the loss of few units or connections. Living systems also display an additional source of fault-tolerance obtained through distributed processes of self-healing: defective components are actively regenerated. In this paper, we present results obtained with a model of development for spiking neural networks undergoing sustained levels of cell loss. To test their resistance to faults, networks are subjected to random faults during development and mutilated several times during operation. Results show that, evolved to control simulated Khepera robots in a simple navigation task, plastic and non-plastic networks develop fault-tolerant structures which can recover normal operation to various degrees.

Algorithms↗
PubMedDate not supplied

Robust nonlinear autoregressive moving average model parameter estimation using stochastic recurrent artificial neural networks.

In this study, we introduce a new approach for estimating linear and nonlinear stochastic autoregressive moving average (ARMA) model parameters, given a corrupt signal, using artificial recurrent neural networks. This new approach is a two-step approach in which the parameters of the deterministic part of the stochastic ARMA model are first estimated via a three-layer artificial neural network (deterministic estimation step) and then reestimated using the prediction error as one of the inputs to the artificial neural networks in an iterative algorithm (stochastic estimation step). The prediction error is obtained by subtracting the corrupt signal of the estimated ARMA model obtained via the deterministic estimation step from the system output response. We present computer simulation examples to show the efficacy of the proposed stochastic recurrent neural network approach in obtaining accurate model predictions. Furthermore, we compare the performance of the new approach to that of the deterministic recurrent neural network approach. Using this simple two-step procedure, we obtain more robust model predictions than with the deterministic recurrent neural network approach despite the presence of significant amounts of either dynamic or measurement noise in the output signal. The comparison between the deterministic and stochastic recurrent neural network approaches is furthered by applying both approaches to experimentally obtained renal blood pressure and flow signals.

Algorithms↗
PubMed1998

Calvarial eosinophilic granuloma: diagnostic models and image feature selection with a neural network.

RATIONALE AND OBJECTIVES: The authors analyzed the accuracy of diagnostic features used by an artificial neural network compared with logistic-regression analysis in the diagnosis with computed tomography (CT) of calvarial eosinophilic granuloma. MATERIALS AND METHODS: Thirty-one of 167 patients with calvarial lesions were found to have eosinophilic granuloma. Clinical and CT data were used for logistic-regression and neural network models. Both models were tested by using the leave-one-out method. The final results of each model were compared by means of the area under the receiver operating characteristic curve (Az). RESULTS: Identification of eosinophilic granuloma was significantly more accurate with the neural network than with logistic regression (Az = 0.9846 +/- 0.0157 [standard deviation] vs 0.9117 +/- 0.0373) (P = .001). The most important diagnostic features identified with the neural network were patient age and marginal sclerosis. For logistic regression, the most important features were age, shape, and lobularity. CONCLUSION: The neural network is a useful tool for analyzing the features of calvarial eosinophilic granuloma. Age and marginal sclerosis are important diagnostic features.

Adolescent↗
PubMed1996

On the Circuit Complexity of Sigmoid Feedforward Neural Networks.

This paper aims to examine the circuit complexity of sigmoid activation feedforward artificial neural networks by placing them amongst several classic Boolean and threshold gate circuit complexity classes. The starting point is the class NN(k) defined by [Shawe-Taylor et al. (1992)] Classes of feedforward neural nets and their circuit complexity. Neural Networks 5(6), 971-977. For a better characterisation, we introduce two additional classes NN(k)(Delta) and NN(k)(Delta,epsilon) having less restrictive conditions than NN(k) concerning fan-in and accuracy, and proceed to prove relations amongst these three classes and well established circuit complexity classes. For doing that, a particular class of Boolean functions F(Delta) is first introduced and we show how a threshold gate circuit can be recursively built for any f(Delta) belonging to F(Delta). As the G-functions (computing the carries) are f(Delta) functions, a class of solutions is obtained for threshold gate adders. We then constructively prove the inclusions amongst circuit complexity classes. This is done by converting the sigmoid feedforward artificial neural network into an equivalent threshold gate circuit [Shawe-Taylor et al. (1992)]. Each threshold gate is then replaced by a multiple input adder having a binary tree structure, relaxing the logarithmic fan-in condition from ([Shawe-Taylor et al. 1992]) to (almost) polynomial. This means that larger classes of sigmoid activation feedforward neural networks can be implemented in polynomial size Boolean circuits with a small constant fan-in at the expense of a logarithmic factor increase in the number of layers. Similar results are obtained for threshold circuits, and are liked with the previous ones. The main conclusion is that there are interesting fan-in dependent depth-size tradeoffs when trying to digitally implement sigmoid activation feedforward neural networks. Copyright 1996 Elsevier Science Ltd

Journal Article↗
PubMed1993

Predicting secondary structures of membrane proteins with neural networks.

Back-propagation, feed-forward neural networks are used to predict the secondary structures of membrane proteins whose structures are known to atomic resolution. These networks are trained on globular proteins and can predict globular protein structures having no homology to those of the training set with correlation coefficients (Ci) of 0.45, 0.32 and 0.43 for alpha-helix, beta-strand and random coil structures, respectively. When tested on membrane proteins, neural networks trained on globular proteins do, on average, correctly predict (Qi) 62%, 38% and 69% of the residues in the alpha-helix, beta-strand and random coil structures. These scores rank higher than those obtained with the currently used statistical methods and are comparable to those obtained with the joint approaches tested so far on membrane proteins. The lower success score for beta-strand as compared to the other structures suggests that the sample of beta-strand patterns contained in the training set is less representative than those of alpha-helix and random coil. Our analysis, which includes the effects of the network parameters and of the structural composition of the training set on the prediction, shows that regular patterns of secondary structures can be successfully extrapolated from globular to membrane proteins.

Databases, Factual↗
PubMed2000

Identifying three-phase induction motor faults using artificial neural networks

This paper presents an artificial neural network (ANN) based technique to identify faults in a three-phase induction motor. The main types of faults considered are overload, single phasing, unbalanced supply voltage, locked rotor, ground fault, over-voltage and under-voltage. Three-phase currents and voltages from the induction motor are used in the proposed approach. A feedforward layered neural network structure is used. The network is trained using the backpropagation algorithm. The trained network is tested with simulated fault current and voltage data. Fault detection is attempted in the no fault to fault transition period. Off-line testing results on a 3 HP induction motor model show that the proposed ANN based method is effective in identifying various types of faults.

Journal Article↗
PubMed1996

Artificial neural networks: a prospective tool for the analysis of psychiatric disorders.

Artificial neural networks are computer simulations of biological parallel distributed processing systems. They are able to undertake complex pattern recognition tasks, including diagnostic classification, prediction of disease onset and prognosis, and identification of determinants of clinical decisions. These capabilities have been utilized in general medicine, but as yet there has been little application of artificial neural networks in psychiatric research. Artificial neural networks can also be used to create models of brain function, providing a paradigm for cognition and the organization of neural systems that demonstrates how changes at the cellular level can affect information processing. These models are able to encompass both the biological and the behavioral dimensions of psychiatric disorders.

Humans↗
PubMed2001

Conjoint and extended neural networks for the computation of speech codes: the neural basis of selective impairment in reading words and pseudowords.

The computation of speech codes (i.e. phonology) is an important aspect of word reading. Understanding the neural systems and mech- anisms underlying phonological processes provides a foundation for the investigation of language in the brain. We used high-resolution three-dimensional positron emission tomography (PET) to investigate neural systems essential for phonological processes. The burden of neural activities on the computation of speech codes was maximized by three rhyming tasks (rhyming words, pseudowords and words printed in mixed letter cases). Brain activation patterns associated with these tasks were compared with those of two baseline tasks involving visual feature detection. Results suggest strong left lateralized epicenters of neural activity in rhyming irrespective of gender. Word rhyming activated the same brain regions engaged in pseudoword rhyming, suggesting conjoint neural networks for phonological processing of words and pseudowords. However, pseudoword rhyming induced the largest change in cerebral blood flow and activated more voxels in the left posterior prefrontal regions and the left inferior occipital-temporal junction. In addition, pseudoword rhyming activated the left supramarginal gyrus, which was not apparent in word rhyming. These results suggest that rhyming pseudowords requires active participation of extended neural systems and networks not observed for rhyming words. The implications of the results on theories and models of visual word reading and on selective reading dysfunctions after brain lesions are discussed.

Adult↗
PubMed2007

Artificial neural network: predicted vs observed survival in patients with colonic cancer.

PURPOSE: An Internet-web-based artificial neural network has been developed for practicing clinical oncologists and medical researchers as part of an ongoing program designed for the implementation of advanced neural networks for prognostic estimates and eventually for management/treatment decisions in individual patients with colonic cancer. METHODS: An interdisciplinary team of academic oncologists and physicists has configured and implemented a Partial Logistic Artificial Neural Network and trained it to predict cancer-related survival in patients with confirmed colorectal cancer by using a database (1,558 patients) made available for the study by the Information & Statistics Division of National Health Service Scotland. The reliability of the trained network was evaluated against Kaplan-Meier observed survival plots of a random sample of 300 patients not used in the training but forming part of the same data set. RESULTS: The predicted survival curves obtained as the output from the artificial neural network showed close agreement with observed actual survival rates of a cohort of 300 patients with four grades of risk of dying from the cancer within five years of diagnosis. CONCLUSIONS: The web-based Partial Logistic Artificial Neural Network system accurately predicts survival after staging and treatment of colonic cancer. It can be made web-accessible where it is powerful enough to serve hundreds of users simultaneously.

Aged↗
PubMed1996

Agreement between artificial neural networks and experienced electrocardiographer on electrocardiographic diagnosis of healed myocardial infarction.

OBJECTIVES: The purpose of this study was to compare the diagnoses of healed myocardial infarction made from the 12-lead electrocardiogram (ECG) by artificial neural networks and an experienced electrocardiographer. BACKGROUND: Artificial neural networks have proved of value in pattern recognition tasks. Studies of their utility in ECG interpretation have shown performance exceeding that of conventional ECG interpretation programs. The latter present verbal statements, often with an indication of the likelihood for a certain diagnosis, such as "possible left ventricular hypertrophy." A neural network presents its output as a numeric value between 0 and 1; however, these values can be interpreted as Bayesian probabilities. METHODS: The study was based on 351 healthy volunteers and 1,313 patients with a history of chest pain who had undergone diagnostic cardiac catheterization. A 12-lead ECG was recorded in each subject. An expert electrocardiographer classified the ECGs in five different groups by estimating the probability of anterior myocardial infarction. Artificial neural networks were trained and tested to diagnose anterior myocardial infarction. The network outputs were divided into five groups by using the output values and four thresholds between 0 and 1. RESULTS: The neural networks diagnosed healed anterior myocardial infarctions at high levels of sensitivity and specificity. The network outputs were transformed to verbal statements, and the agreement between these probability estimates and those of an expert electrocardiographer was high. CONCLUSIONS: Artificial neural networks can be of value in automated interpretation of ECGs in the near future.

Electrocardiography↗
PubMed2001

A closed-form neural network for discriminatory feature extraction from high-dimensional data.

We consider a new neural network for data discrimination in pattern recognition applications. We refer to this as a maximum discriminating feature (MDF) neural network. Its weights are obtained in closed-form, thereby overcoming problems associated with other nonlinear neural networks. It uses neuron activation functions that are dynamically chosen based on the application. It is theoretically shown to provide nonlinear transforms of the input data that are more general than those provided by other nonlinear multilayer perceptron neural network and support-vector machine techniques for cases involving high-dimensional (image) inputs where training data are limited and the classes are not linearly separable. We experimentally verify this on synthetic examples.

Algorithms↗
PubMed2005

Artificial neural network models for prediction of acute coronary syndromes using clinical data from the time of presentation.

STUDY OBJECTIVE: Clinical and ECG data from presentation are highly discriminatory for diagnosis of acute coronary syndromes, whereas definitive diagnosis from serial ECG and cardiac marker protein measurements is usually not available for several hours. Artificial neural networks are computer programs adept at pattern recognition tasks and have been used to analyze data from chest pain patients with a view to developing diagnostic algorithms that might improve triage practices in the emergency department. The aim of this study is to develop and optimize artificial neural network models for diagnosis of acute coronary syndrome, to test these models on data collected prospectively from different centers, and to establish whether the performance of these models was superior to that of models derived using a standard statistical technique, logistic regression. METHODS: The study used data from 3,147 patients presenting to 3 hospitals with acute chest pain. Data from hospital 1 were used to train the models, which were then tested on independent data from the other 2 hospitals. From 40 potential factors, variables were selected according to the logarithm of their likelihood ratios to produce models using 8, 13, 20, and 40 factors. Identical data were used for logistic regression and artificial neural network models. Calibration and performance were assessed, the latter using receiver operating characteristic (ROC) curve analysis. RESULTS: Although the performance of artificial neural network models generally increased with increasing numbers of factors, this was insignificant. The 13-factor model was therefore used for the rest of the study owing to its marginally improved calibration over the smallest model. Area under the ROC curve (with standard error) was 0.97 (0.006). The overall sensitivity and specificity of this model for acute coronary syndrome diagnosis using the training data was 0.93. ROC curves for logistic regression and artificial neural network models applied to data from the 3 hospitals were identical. For the 13-factor artificial neural network model tested on data from hospitals 2 and 3, area under the ROC curves (standard error) were 0.93 (0.006) and 0.95 (0.009), respectively. Investigation of the performance of the artificial neural network models throughout the range of predicted probabilities showed that they were well calibrated. CONCLUSION: This study confirms that artificial neural networks can offer a useful approach for developing diagnostic algorithms for chest pain patients; however, the exceptional performance and simplicity of the logistic model militates in favor of logistic regression for the present task. Our artificial neural network models were well calibrated and performed well on unseen data from different centers. These issues have not been addressed in previous studies. However, and unlike in previous studies, we did not find the performance of artificial neural network models to be significantly different from that of suitably optimized logistic regression models.

Adolescent↗
PubMed2002

Usefulness of an artificial neural network for differentiating benign from malignant pulmonary nodules on high-resolution CT: evaluation with receiver operating characteristic analysis.

OBJECTIVE: The purpose of our study was to use an artificial neural network to differentiate benign from malignant pulmonary nodules on high-resolution CT findings and to evaluate the effect of artificial neural network output on the performance of radiologists using receiver operating characteristic analysis. MATERIALS AND METHODS: We selected 155 cases with pulmonary nodules less than 3 cm (99 malignant nodules and 56 benign nodules). An artificial neural network was used to distinguish benign from malignant nodules on the basis of seven clinical parameters and 16 radiologic findings that were extracted by attending radiologists using subjective rating scales. In the observer test, 12 radiologists (four attending radiologists, four radiology fellows, and four radiology residents) were presented with high-resolution CT images, first without and then with the artificial neural network output. Observer performance was evaluated by means of receiver operating characteristic analysis using a continuous rating scale. RESULTS: The artificial neural network showed a high performance in differentiating benign from malignant pulmonary nodules (A(z) = 0.951). The average A(z) value for all radiologists increased by a statistically significant level, from 0.831 to 0.959, with the use of the artificial neural network output. CONCLUSION: Our computerized scheme using the artificial neural network can improve the diagnostic accuracy of radiologists who are differentiating benign from malignant pulmonary nodules on high-resolution CT.

Adult↗
PubMed1999

Artificial neural networks in urology: Update 2000.

Artificial neural networks (ANNs) are widely available and have been demonstrated to be superior to standard empirical methods of detecting, staging and monitoring prostate cancer. These algorithms have been statistically validated in diverse, well-characterized patient groups and are now being evaluated for clinical use worldwide. New variables based on demographic data, tissue and serum markers show promise for improving our ability to predict disease extent and outcome and may be integrated in future ANN models. This review focuses on recently developed neural networks for detecting, staging and monitoring prostate cancer.

Journal Article↗
PubMed2001

Comparison of the prediction of extremely low birth weight neonatal mortality by regression analysis and by neural networks.

AIMS: To compare the prediction of mortality in individual extremely low birth weight (ELBW) neonates by regression analysis and by artificial neural networks. STUDY DESIGN: A database of 23 variables on 810 ELBW neonates admitted to a tertiary care center was divided into training, validation, and test sets. Logistic regression and neural network models were developed on the training set, validated, and outcome (mortality) predicted on the test set. Stepwise regression identified significant variables in the full set. Regression models and neural networks were then tested using data sets with only the identified significant variables, and then with variables excluded one at a time. RESULTS: The area under the curve (AUC) of receiver operating characteristic (ROC) curves for neural networks and regression was similar (AUC 0.87+/-0.03; p=0.31). Birthweight or gestational age and the 5-min Apgar score contributed most to AUC. CONCLUSIONS: Both neural networks and regression analysis predicted mortality with reasonable accuracy. For both models, analyzing selected variables was superior to full data set analysis. We speculate neural networks may not be superior to regression when no clear non-linear relationships exist.

Area Under Curve↗
PubMed1995

Detection of compositional constraints in nucleic acid sequences using neural networks.

We describe in this paper a neural network method for the detection of compositional constraints in introns and exons. The first part of the algorithm (learning phase) consisted in presenting examples of intron and exon sequences to the network and in modifying its connections using the back-propagation algorithm. Previous connectionist methods achieved the learning of exons and introns using the latter as negative examples to the former. However, we chose to learn introns and exons jointly, using junk DNA as a common counter-example. In a second part (generalization phase), we tested the neural networks in the search for exons and introns in the human globin cluster. Their performances were also checked on the classification of unknown examples. As with the previous approaches, this technique discriminates introns and exons: values of the correlation coefficients are respectively 0.50 and 0.64 for the best achieved network. Moreover, using junk DNA sequences in the learning phase allows one to detect constrained regions inside the intron and the exon sequences (i.e. sequences that differ, by their nucleic acid compositions, from junk DNA). The application of our approach could be useful in the study of the internal organization of these sequences.

Algorithms↗
PubMed2001

ECG beat classification by a novel hybrid neural network.

This paper presents a novel hybrid neural network structure for the classification of the electrocardiogram (ECG) beats. Two feature extraction methods: Fourier and wavelet analyses for ECG beat classification are comparatively investigated in eight-dimensional feature space. ECG features are determined by dynamic programming according to the divergence value. Classification performance, training time and the number of nodes of the multi-layer perceptron (MLP), restricted Coulomb energy (RCE) and a novel hybrid neural network are comparatively presented. In order to increase the classification performance and to decrease the number of nodes, the novel hybrid structure is trained by the genetic algorithms (GAs). Ten types of ECG beats obtained from the MIT-BIH database and from a real-time ECG measurement system are classified with a success of 96% by using the hybrid structure.

Algorithms↗