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.

SourceResearch connectionFrom yearThrough year

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

At least 415 records · Page 23Linked to original sources

PubMed2002

Selecting neural networks for a committee decision.

To improve recognition results, decisions of multiple neural networks can be aggregated into a committee decision. In contrast to the ordinary approach of utilizing all neural networks available to make a committee decision, we propose creating adaptive committees, which are specific for each input data point. A prediction network is used to identify classification neural networks to be fused for making a committee decision about a given input data point. The jth output value of the prediction network expresses the expectation level that the jth classification neural network will make a correct decision about the class label of a given input data point. The proposed technique is tested in three aggregation schemes, namely majority vote, averaging, and aggregation by the median rule and compared with the ordinary neural networks fusion approach. The effectiveness of the approach is demonstrated on two artificial and three real data sets.

Artificial Intelligence↗

PubMed2005

An efficient parameterization of dynamic neural networks for nonlinear system identification.

Dynamic neural networks (DNNs), which are also known as recurrent neural networks, are often used for nonlinear system identification. The main contribution of this letter is the introduction of an efficient parameterization of a class of DNNs. Having to adjust less parameters simplifies the training problem and leads to more parsimonious models. The parameterization is based on approximation theory dealing with the ability of a class of DNNs to approximate finite trajectories of nonautonomous systems. The use of the proposed parameterization is illustrated through a numerical example, using data from a nonlinear model of a magnetic levitation system.

Algorithms↗

PubMed1994

Model reference direct adaptive control of nonlinear plants using neural networks.

A learning scheme for multilayer feedforward neural networks used as direct adaptive controllers of nonlinear plants is suggested. This scheme is a supervised steepest descent one that does not require backpropagation of the error. Using a neural network controller trained with this method does not require the identification stage and this makes it superior to the other methodologies. Methods for using neural networks in plant control suggested in the literature are discussed and compared with the proposed system. The structure of the network and the training method used are explained. Simulations based on model reference control of some nonlinear plants show satisfactory performance.

Computer Simulation↗

PubMed2004

Integration of clinical and imaging data to predict the presence of coronary artery disease with the use of neural networks.

AIM: To establish proof of the principle that a computer-based neural network method can be employed that will enhance diagnostic accuracy vis-a-vis image analysis alone in the interpretation of treadmill exercise tests performed in conjunction with myocardial perfusion imaging. MATERIALS AND METHODS: One-hundred-and-two patients underwent myocardial perfusion imaging in association with the standard Bruce protocol. Twenty objective parameters describing each patient's exercise physiology, general clinical status and image appearance were used to train an artificial neural network. Classification accuracy of the neural network and clinical interpretation was determined by coronary angiography. We evaluated the ability of the neural network to integrate clinical, exercise and imaging data to determine the likelihood of coronary artery disease and compared these results with an optimized method of clinical image interpretation, which made use of all available clinical, angiographic and stress test data. RESULTS: The artificial neural network had a sensitivity of 88% and a specificity of 65% for detection of ischemic heart disease and was comparable to that of the optimized clinical method (sensitivity 80%, specificity 69%). Incorporation of clinical and exercise data significantly improved the predictive accuracy of the network compared to a network based on image data alone (P<0.05). CONCLUSION: The results show a computer-based neural network can perform as well as expert readers working under optimal conditions including full knowledge of the patient's clinical, prior angiographic and stress test data. Thus, the method is promising as a diagnostic aid to the recognition of ischemic heart disease in the clinical setting of treadmill exercise testing in conjunction with myocardial perfusion imaging.

Aged↗

PubMed2006

Tuning the structure and parameters of a neural network by using hybrid Taguchi-genetic algorithm.

In this paper, a hybrid Taguchi-genetic algorithm (HTGA) is applied to solve the problem of tuning both network structure and parameters of a feedforward neural network. The HTGA approach is a method of combining the traditional genetic algorithm (TGA), which has a powerful global exploration capability, with the Taguchi method, which can exploit the optimum offspring. The Taguchi method is inserted between crossover and mutation operations of a TGA. Then, the systematic reasoning ability of the Taguchi method is incorporated in the crossover operations to select the better genes to achieve crossover, and consequently enhance the genetic algorithms. Therefore, the HTGA approach can be more robust, statistically sound, and quickly convergent. First, the authors evaluate the performance of the presented HTGA approach by studying some global numerical optimization problems. Then, the presented HTGA approach is effectively applied to solve three examples on forecasting the sunspot numbers, tuning the associative memory, and solving the XOR problem. The numbers of hidden nodes and the links of the feedforward neural network are chosen by increasing them from small numbers until the learning performance is good enough. As a result, a partially connected feedforward neural network can be obtained after tuning. This implies that the cost of implementation of the neural network can be reduced. In these studied problems of tuning both network structure and parameters of a feedforward neural network, there are many parameters and numerous local optima so that these studied problems are challenging enough for evaluating the performances of any proposed GA-based approaches. The computational experiments show that the presented HTGA approach can obtain better results than the existing method reported recently in the literature.

Algorithms↗

PubMed1990

Potential usefulness of an artificial neural network for differential diagnosis of interstitial lung diseases: pilot study.

An artificial neural network approach was applied to the differential diagnosis of interstitial lung diseases. The neural network was designed to distinguish between nine types of interstitial lung diseases on the basis of 20 items of clinical and radiographic information. A data base for training and testing the neural network was created with 10 hypothetical cases for each of the nine diseases. The performance of the neural network was evaluated by means of receiver operating characteristic analysis. The decision performance of the neural network was high; it was comparable to that of chest radiologists and superior to that of senior radiology residents. The preliminary results strongly suggest that the neural network approach has potential utility in the computer-aided differential diagnosis of interstitial lung diseases.

Computer Simulation↗

PubMedDate not supplied

Predicting mortality after coronary artery bypass surgery: what do artificial neural networks learn? The Steering Committee of the Cardiac Care Network of Ontario.

OBJECTIVE: To compare the abilities of artificial neural network and logistic regression models to predict the risk of in-hospital mortality after coronary artery bypass graft (CABG) surgery. METHODS: Neural network and logistic regression models were developed using a training set of 4,782 patients undergoing CABG surgery in Ontario, Canada, in 1991, and they were validated in two test sets of 5,309 and 5,517 patients having CABG surgery in 1992 and 1993, respectively. RESULTS: The probabilities predicted from a fully trained neural network were similar to those of a "saturated" regression model, with both models detecting all possible interactions in the training set and validating poorly in the two test sets. A second neural network was developed by cross-validating a network against a new set of data and terminating network training early to create a more generalizable model. A simple "main effects" regression model without any interaction terms was also developed. Both of these models validated well, with areas under the receiver operating characteristic curves of 0.78 and 0.77 (p > 0.10) in the 1993 test set. The predictions from the two models were very highly correlated (r=0.95). CONCLUSIONS: Artificial neural networks and logistic regression models learn similar relationships between patient characteristics and mortality after CABG surgery.

Aged↗

PubMed1998

A multilayer recurrent neural network for solving continuous-time algebraic Riccati equations.

A multilayer recurrent neural network is proposed for solving continuous-time algebraic matrix Riccati equations in real time. The proposed recurrent neural network consists of four bidirectionally connected layers. Each layer consists of an array of neurons. The proposed recurrent neural network is shown to be capable of solving algebraic Riccati equations and synthesizing linear-quadratic control systems in real time. Analytical results on stability of the recurrent neural network and solvability of algebraic Riccati equations by use of the recurrent neural network are discussed. The operating characteristics of the recurrent neural network are also demonstrated through three illustrative examples.

Journal Article↗

PubMed2006

Prostate cancer nomograms are superior to neural networks.

INTRODUCTION: Several nomograms have been developed to predict PCa related outcomes. Neural networks represent an alternative. METHODS: We provide a descriptive and an analytic comparison of nomograms and neural networks, with focus on PCa detection. RESULTS: Our results indicate that nomograms have several advantages that distinguish them from neural networks. These are both quantitative and qualitative. CONCLUSION: In the field of PCa detection, nomograms appear to outweigh the benefits of neural networks. However, the neural network methodology represents a valid alternative, which should not be underestimated.

Forecasting↗

PubMed1999

Inferential estimation of polymer quality using bootstrap aggregated neural networks.

Inferential estimation of polymer quality in a batch polymerisation reactor using bootstrap aggregated neural networks is studied in this paper. Number average molecular weight and weight average molecular weight are estimated from the on-line measurements of reactor temperature, jacket inlet and outlet temperatures, coolant flow rate through the jacket, monomer conversion, and the initial batch conditions. Bootstrap aggregated neural networks are used to enhance the accuracy and robustness of neural network models built from a limited amount of training data. The training data set is re-sampled using bootstrap re-sampling with replacement to form several sets of training data. For each set of training data, a neural network model is developed. The individual neural networks are then combined together to form a bootstrap aggregated neural network. Determination of appropriate weights for combining individual networks using principal component regression is proposed in this paper. Confidence bounds for neural network predictions can also be obtained using the bootstrapping technique. The techniques have been successfully applied to the simulation of a batch methyl methacrylate polymerisation reactor.

Journal Article↗

PubMed1992

A parallel implementation of the backward error propagation neural network training algorithm: experiments in event identification.

An artificial neural-network-based (ANN) event detection and alarm generation system has been developed to aid clinicians in the identification of critical events commonly occurring in the anesthesia breathing circuit. To detect breathing circuit problems, the system monitored CO2 gas concentration, gas flow, and airway pressure. Various parameters were extracted from each of these input waveforms and fed into an artificial neural network. To develop truly robust ANNs, investigators are required to train their networks on large training data sets, requiring enormous computing power. We implemented a parallel version of the backward error propagation neural network training algorithm in the widely portable parallel programming language C-Linda. A maximum speedup of 4.06 was obtained with six processors. This speedup represents a reduction in total run-time from 6.4 to 1.5 h. By reducing the total run time of the computation through parallelism, we were able to optimize many of the neural network's initial parameters. We conclude that use of the master-worker model of parallel computation is an excellent method for speeding up the backward error propagation neural network training algorithm.

Algorithms↗

PubMed1997

A study of learning splice sites of DNA sequence by neural networks.

To predict of splice sites in DNA sequence, we developed a neural network system with back propagation. This system has a flexible network definition language which can describe any network structure. Three types of neural network were defined using the system for the prediction of splice sites. The neural networks are trained by the arrangements of bases around the splice sites of DNA sequences. The results of simulation showed the excellent ability of the neural networks to predict splice sites by applying and testing the arrangements of DNA sequences. This system also were used to predict the effects of point mutations on the splicing of the IX factor gene which may cause hereditary disease.

Algorithms↗

PubMed2004

Prediction of disulfide-bonded cysteines in proteomes with a hidden neural network.

A hidden neural network-based method is used to predict the bonding state of cysteines starting from the residue sequence of the protein chain. The method scores as high as 89% and 86% per cysteine residue and per protein, respectively, and in this overcomes other predictors of the same category. We then explore the efficacy of our predictor in computing the disulfide content of the whole proteome of Escherichia coli (K12 and O157), Aeropirum pernix, Thermotoga maritima, and Homo sapiens. We find that the percentage of extracellular disulfide containing proteins is higher than that of intracellular one, and that the human proteome is by far the one with the highest content of sulfur-sulfur linkages in proteins.

Cysteine↗

PubMedDate not supplied

Neural networks in the study of the brain.

Neural networks are models of the brain and have been used within Artificial Intelligence to provide alternative explanations to the symbolic explanations of cognition in which one assumes that an intelligent system has certain explicit representation of some aspect of the world and uses these in intelligent behavior. Obviously, if neural networks are indeed good models of the brain, and give a satisfactory account of cognition, then they could be a valuable tool to neuroscientists. This article gives a brief overview of the various neural network models, and critically reviews their status as models of the brain and of cognition.

Animals↗

PubMed1998

Neural network computer program to determine photorefractive keratectomy nomograms.

PURPOSE: To evaluate a commercially available neural network program for calculation of photorefractive keratectomy treatment nomograms. SETTING: University referral refractive surgery clinic. METHODS: PRK/LASIK Brain, a commercial neural network computer program, was trained using the demographics, preoperative clinical data, surgical parameters, and 1 year postoperative clinical data of 44 patients treated with a Summit Technology excimer laser using a 5.0 mm optical zone. The neural-network derived nomogram was compared with the standard treatment nomogram for each patient. The relative contribution of age, sex, keratometry, and intraocular pressure (IOP) to the predicted nomograms was also assessed. RESULTS: Nomograms produced by the neural network were qualitatively similar to the standard nomogram. The sequence of data entry during training affected the network's predictions. Entry ordered by outcome (as opposed to entry by chronological order) yielded a nomogram that was more consistent with the standard nomogram. However, both outcome- and chronologically ordered network-derived nomograms diverged from the standard nomogram in individual patients, including a subset for whom use of the standard nomogram yielded desired refractive results (within 0.25 diopter of emmetropia). Further analysis of the neural-network-derived nomograms revealed marked sensitivity to sex, age, keratometry readings, and IOP. CONCLUSIONS: Neural networks offer a potential means of individualizing treatment nomograms, to account for patient demographics, preoperative examination, surgeon style, and equipment bias. However, a data set of 44 patients was not sufficient to train the PRK/LASIK Brain network to accurately predict treatment parameters in individual cases in the training set. A larger training set or a different learning algorithm may be required to improve the neural network's performance.

Adult↗

PubMed1994

Diagnosis of focal bone lesions using neural networks.

RATIONALE AND OBJECTIVES: Use of a neural network to diagnose focal lesions of bone was evaluated. METHODS: Imaging features of 709 lesions were encoded into a predetermined database. Data were divided into four groups and were analyzed using cross-validation by a two-layer feed-forward neural network. RESULTS: The lesions comprised 43 different pathologic diagnoses. Overall, the network was 85% accurate in distinguishing benign from malignant lesions. With a differential list of five diagnoses, the list was internally consistent regarding benign and malignant lesions 81.9% of the time. The network correctly diagnosed 56% of the lesions by pathologic diagnosis as its first choice. It included the correct diagnosis 71.8% of the time in a differential list of three diagnoses and 87.3% of the time in a differential list of nine diagnoses. CONCLUSION: Although not yet adequate for clinical use, neural network diagnosis of bone lesions is in its infancy and has important implications for the future analysis of focal bone lesions.

Bone Neoplasms↗

PubMed1997

Transformation of sensory signals into commands for saccadic eye movements: a neural network study.

A biological plausible neural network which simulated the input-output transformation performed by primates during saccadic eye movements is constructed using a selective attention module and multi-layered neural networks with improved back propagation and a competitive learning algorithm. Simulation results show that the trained model can make fine saccades directed by the target. Representations and processing mechanisms in the saccade system are investigated. The features of most hidden units resemble those that have been observed in physiological recordings of neurons in primates visual cortex. The hidden layer even developed structures similar to those of area 7a.

Algorithms↗

PubMed1993

Artificial neural networks as adjuncts for assessing medical students' problem solving performances on computer-based simulations.

Artificial neural networks were trained by supervised learning to recognize the test selection patterns associated with students' successful solutions to seven immunology computer-based simulations. New test selection patterns evaluated by the trained neural network were correctly classified as successful or unsuccessful solutions to the problem > 90% of the time. The examination of the neural networks output weights after each test selection revealed a progressive and selective increase for the relevant problem suggesting that a successful solution is represented by the neural network as the accumulation of relevant tests. Unsuccessful problem solutions were classified by the neural network software into two patterns of students performance. The first pattern was characterized by low neural network output weights for all seven problems reflecting extensive searching and lack of recognition of relevant information. In the second pattern, the output weights from the neural network were biased toward one of the remaining six incorrect problems suggesting that the student misrepresented the current problem as an instance of a previous problem. Finally, neural network analysis could detect cases where the students switched hypotheses during the problem solving exercises.

California↗