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 181 records · Page 10Linked to original sources
PubMed1988

Recognition of general patterns using neural networks.

The Hopfield model of neural network stores memory in its symmetric synaptic connections and can only learn to recognize sets of nearly "orthogonal" patterns. A new algorithm is put forth to permit the recognition of general ("non-orthogonal") patterns. The algorithm specifies the construction of the new network's memory matrix Tij, which is, in general, asymmetrical and contains the Hopfield neural network (Hopfield 1982) as a special case. We find further that in addition to this new algorithm for general pattern recognition, there exists in fact a large class of Tij memory matrices which permit the recognition of non-orthogonal patterns. The general form of this class of Tij memory matrix is presented, and the projection matrix neural network (Personnaz et al. 1985) is found as a special case of this general form. This general form of memory matrix extends the library of memory matrices which allow a neural network to recognize non-orthogonal patterns. A neural network which followed this general form of memory matrix was modeled on a computer and successfully recognized a set of non-orthogonal patterns. The new network also showed a tolerance for altered and incomplete data. Through this new method, general patterns may be taught to the neural network.

Algorithms↗
PubMedDate not supplied

Predicting the duration of the first stage of spontaneous labor using a neural network.

To create a neural network that predicts the length of the first stage of term labor. Two hundred patients with gestations > or = 36 weeks, in spontaneous active labor are the study group: 159 for training and 41 for testing; 4 training set patients had second-stage cesarean section for obstructed labor. The network is designed with Brainmaker MacIntosh 1.0 (California Scientific Software). Inputs are uterine activity, estimated fetal weight, position, station, and gestational age; maternal parity, age, height, weight, membrane status, and cervical dilatation. Actual first stages are regressed on those predicted by the network or by a standard partogram set. Differences between actual first stage lengths and those predicted by the neural network or partogram are compared with t-tests; while the proportions of first stages accurately predicted within 1 or 2 h are compared for both methods with chi-square tests. The network trained in 4 h (1388 runs) to a 0.15 tolerance. The network predictions have significantly higher correlation (r = 0.88) than do standard partograms (r = 0.35) with actual first stage durations. Mean differences between predicted and actual first stages are significantly lower for network output than with partograms; these differences increased with first stages exceeding 3 h; 100% of trained network values are within 2 h of actual first stage length. The network performs similarly for a new set of 41 previously unseen labors. This neural network predicts the length of the first stage of spontaneous labor and uses inputs readily available to obstetricians. It outperforms typical partograms for estimating this important feature of normal labor. Future application for intrapartum prognosis could be based on this successful design.

Artificial Intelligence↗
PubMed2001

A real-coded genetic algorithm for training recurrent neural networks.

The use of Recurrent Neural Networks is not as extensive as Feedforward Neural Networks. Training algorithms for Recurrent Neural Networks, based on the error gradient, are very unstable in their search for a minimum and require much computational time when the number of neurons is high. The problems surrounding the application of these methods have driven us to develop new training tools. In this paper, we present a Real-Coded Genetic Algorithm that uses the appropriate operators for this encoding type to train Recurrent Neural Networks. We describe the algorithm and we also experimentally compare our Genetic Algorithm with the Real-Time Recurrent Learning algorithm to perform the fuzzy grammatical inference.

Algorithms↗
PubMed1996

Detection of seizure activity in EEG by an artificial neural network: a preliminary study.

Neural networks, inspired by the organizational principles of the human brain, have recently been used in various fields of application such as pattern recognition, identification, classification, speech, vision, signal processing, and control systems. In this study, a two-layered neural network has been trained for the recognition of temporal patterns of the electroencephalogram (EEG). This network is called a Learning Vector Quantization (LVQ) neural network since it learns the characteristics of the signal presented to it as a vector. The first layer is a competitive layer which learns to classify the input vectors. The second, linear, layer transforms the output of the competitive layer to target classes defined by the user. We have tested and evaluated the LVQ network. The network successfully detects epileptiform discharges (EDs) when trained using EEG records scored by a neurologist. Epochs of EEG containing EDs from one subject have been used for training the network, and EEGs of other subjects have been used for testing the network. The results demonstrate that the LVQ detector can generalize the learning to previously "unseen" records of subjects. This study shows that the LVQ network offers a practical solution for ED detection which is easily adjusted to an individual neurologist's style and is as sensitive and specific as an expert visual analysis.

Adolescent↗
PubMed1997

Automatic detection of glaucomatous visual field progression with neural networks.

OBJECTIVE: To evaluate computerized neural networks to determine visual field progression in patients with glaucoma. METHODS: Two hundred thirty-three series of Octopus G1 visual fields of 181 patients with glaucoma were collected. Each series was composed of 4 or more reliable visual fields from patients who had previously undergone automated perimetry. The visual fields were independently evaluated in a masked fashion by 3 experienced observers (K.N.-M, M.W., and J.C.) and were judged to show progression based on the agreement of 2 observers. The stable and progressed series were matched for mean defect at baseline. The threshold data were submitted to a back propagation neural network that was trained to classify each series as stable or progressed. Two thirds of the data were used for the training and the remaining one third to test the performance of the network. This was repeated 3 times to classify all of the series (changing the training and test series). RESULTS: Fifty-nine series of visual fields showed progression and 151 were judged stable. Neural network sensitivity was 73% and specificity was 88% (threshold for progression = 0.5). The concordance of the neural network with the observers was good (0.50 < or = kappa > or = 0.64). CONCLUSIONS: A neural network can be trained to recognize visual field progression in good concordance with experienced observers. Neural networks may be used to aid the physician in the evaluation of glaucomatous visual field progression.

Aged↗
PubMed2005

Simultaneous perturbation learning rule for recurrent neural networks and its FPGA implementation.

Recurrent neural networks have interesting properties and can handle dynamic information processing unlike ordinary feedforward neural networks. However, they are generally difficult to use because there is no convenient learning scheme. In this paper, a recursive learning scheme for recurrent neural networks using the simultaneous perturbation method is described. The detailed procedure of the scheme for recurrent neural networks is explained. Unlike ordinary correlation learning, this method is applicable to analog learning and the learning of oscillatory solutions of recurrent neural networks. Moreover, as a typical example of recurrent neural networks, we consider the hardware implementation of Hopfield neural networks using a field-programmable gate array (FPGA). The details of the implementation are described. Two examples of a Hopfield neural network system for analog and oscillatory targets are shown. These results show that the learning scheme proposed here is feasible.

Algorithms↗
PubMedDate not supplied

A survey of partially connected neural networks.

Almost all artificial neural networks are by default fully connected, which often implies a high redundancy and complexity. Little research has been devoted to the study of partially connected neural networks, despite its potential advantages like reduced training and recall time, improved generalization capabilities, reduced hardware requirements, as well as being a step closer to biological reality. This publication presents an extensive survey of the various kinds of partially connected neural networks, clustered into a clear framework, followed by a detailed comparative discussion.

Artificial Intelligence↗
PubMed1992

Combining logistic regression and neural networks to create predictive models.

Neural networks are being used widely in medicine and other areas to create predictive models from data. The statistical method that most closely parallels neural networks is logistic regression. This paper outlines some ways in which neural networks and logistic regression are similar, shows how a small modification of logistic regression can be used in the training of neural network models, and illustrates the use of this modification for variable selection and predictive model building with neural networks.

Algorithms↗
PubMed2006

Noise in genetic and neural networks.

Both neural and genetic networks are significantly noisy, and stochastic effects in both cases ultimately arise from molecular events. Nevertheless, a gulf exists between the two fields, with researchers in one often being unaware of similar work in the other. In this Special Issue, we focus on bridging this gap and present a collection of papers from both fields together. For each field, the networks studied range from just a single gene or neuron to endogenous networks. In this introductory article, we describe the sources of noise in both genetic and neural systems. We discuss the modeling techniques in each area and point out similarities. We hope that, by reading both sets of papers, ideas developed in one field will give insight to scientists from the other and that a common language and methodology will develop.

Animals↗
PubMed2002

On the emergence of rules in neural networks.

A simple associationist neural network learns to factor abstract rules (i.e., grammars) from sequences of arbitrary input symbols by inventing abstract representations that accommodate unseen symbol sets as well as unseen but similar grammars. The neural network is shown to have the ability to transfer grammatical knowledge to both new symbol vocabularies and new grammars. Analysis of the state-space shows that the network learns generalized abstract structures of the input and is not simply memorizing the input strings. These representations are context sensitive, hierarchical, and based on the state variable of the finite-state machines that the neural network has learned. Generalization to new symbol sets or grammars arises from the spatial nature of the internal representations used by the network, allowing new symbol sets to be encoded close to symbol sets that have already been learned in the hidden unit space of the network. The results are counter to the arguments that learning algorithms based on weight adaptation after each exemplar presentation (such as the long term potentiation found in the mammalian nervous system) cannot in principle extract symbolic knowledge from positive examples as prescribed by prevailing human linguistic theory and evolutionary psychology.

Algorithms↗
PubMed2000

Exploring Alternative Models of Complex Patient Management with Artificial Neural Networks.

This study applied an unsupervised neural network modeling process to test data of the National Board of Medical Examiners (NBME) Computer-based Clinical Scenarios (CCS) to identify new performance categories and validate this process as a scoring technique. The classifications resulting from this neural network modeling were consistent with the NBME model in that highly rated NMBE performances (ratings of 7 or 8) were clustered together on the neural network output grid. Very low performance ratings appeared to share few common features and were accordingly classified at isolated nodes. This clustering was reproducible across three separately trained networks with greater than 80% agreement in two of the three networks trained. However, the neural network also contained performance clusters where disparate NBME-based ratings ranged from 1 (worst) to 8 (best). Here, agreement between networks was less than 60%. Through visualization of the search strategies (search path mapping), this neural network clustering was found to be sensitive to quantitative and qualitative test selections such as excessive usage of irrelevant tests reflecting broader behavioral classification in some instances. A disparity between NBME ratings and an independent human rating system was detected by the neural network model since disagreement among raters was also reflected by a lack of neural network performance clustering. Agreement between rating systems, however, was correlated with neural network clustering for 92% of the highly rated performances.

Journal Article↗
PubMed1993

Application of neural networks to pharmacodynamics.

Neural networks (NN) are computational systems implemented in software or hardware that attempt to simulate the neurological processing abilities of biological systems. A synopsis is presented of the operational characteristics, structures, and applications of NN. The NN technology has primarily been aimed at recognition science (e.g., handwriting, voice, signal, picture, image, pattern, etc.). It is pointed out that NN may also be particularly suitable to deal with pharmacokinetic (PK) and pharmacodynamic (PD) systems, especially in cases such as multivariate PK/PD population kinetics when the systems are so complex that modeling by a conventional structured model building technique is very troublesome. The main practical advantage of NN is the intrinsic ability to closely emulate virtually any multivariate system, including nonlinear systems, independently of structural/physiologic relevance. Thus, NN are most suitable to model the behavior of complex kinetic systems and unsuitable to model the structure. In a practical sense, this structure limitation may be inconsequential because NN in its multivariate formulation may consider any physiologic, clinical, or population variable that may influence the kinetic behavior. The application of NN in PD is demonstrated in terms of the ability of an NN to predict, by extrapolation, the central nervous system (CNS) activity of alfentanil. The drug was infused by a complex computer-controlled infusion scheme over 180 min with simultaneous recording of the CNS effect quantified by a fast Fourier transform power spectrum analysis. The NN was trained to recognize (emulate) the drug input-drug effect behavior of the PD system with the input-effect data for the 180 min as a training set.(ABSTRACT TRUNCATED AT 250 WORDS)

Alfentanil↗
PubMed1996

Hardware prototypes of a Boolean neural network and the simulated annealing optimization method.

Boolean Neural Network is a neural network that operates with binary weight values of "1" and "0". Otherwise it is formally analogous to the Multilayer Perceptron (MLP). Simulated Annealing is a stochastic optimization methods that is suitable for performing nonlinear multivariable optimization tasks. Training a Boolean Neural Network is a well-suited problem to this algorithm. However, the Simulated Annealing method is computationally heavy, which makes the training procedure slow. The training speed can be improved by using custom designed hardware for the whole system including the optimization method and the neural network. Hardware prototypes of a Boolean Neural Network and the Simulated Annealing optimization method have been designed using discrete components. The Boolean Neural Network implementation is basically a dynamically configurable feedforward network of Boolean logic gates of two inputs. The Simulated Annealing implementation is a general purpose hardware tool for multivariable optimization tasks. Here it is applied to do supervised training of the Boolean Neural Network hardware.

Algorithms↗
PubMed1997

Simplex Memory Neural Networks.

The biological neural network-simplex memory neural network-is proposed to describe the mechanisms of pattern memory in the brain. A mathematical model of the simplex memory neural network is constructed to memorize any binary pattern with content-addressable memory function. Under Hebbian learning rule, the new network has some important functions in accord with the learning and memory behaviors of the brain. Copyright 1996 Elsevier Science Ltd.

Journal Article↗
PubMed1995

Hierarchical neural networks for survival analysis.

Neural networks offer the potential of providing more accurate predictions of survival time than do traditional methods. Their use in medical applications has, however, been limited, especially when some data is censored or the frequency of events is low. To reduce the effect of these problems, we have developed a hierarchical architecture of neural networks that predicts survival in a stepwise manner. Predictions are made for the first time interval, then for the second, and so on. The system produces a survival estimate for patients at each interval, given relevant covariates, and is able to handle continuous and discrete variables, as well as censored data. We compared the hierarchical system of neural networks with a nonhierarchical system for a data set of 428 AIDS patients. The hierarchical model predicted survival more accurately than did the nonhierarchical (although both had low sensitivity). The hierarchical model could also learn the same patterns in less than half the time required by the nonhierarchical model. These results suggest that the use of hierarchical systems is advantageous when censored data is present, the number of events is small, and time-dependent variables are necessary.

Acquired Immunodeficiency Syndrome↗
PubMed1995

Solitary pulmonary nodules: determining the likelihood of malignancy with neural network analysis.

PURPOSE: To test a neural network in differentiation of benign from malignant solitary pulmonary nodules. MATERIALS AND METHODS: Neural networks were trained and tested on the characteristics of 318 nodules. Predictive accuracy of the network was judged for calibration and discrimination. Network results were compared with those with a simpler Bayesian method. RESULTS: The Brier score was 0.142 (calibration, 0.003; discrimination, 0.139) for the neural network and 0.133 for the Bayesian analysis (calibration, 0.012; discrimination, 0.121). Analysis of the calibration curve revealed no significant difference (P < .05) between the slope (b = 1.09) and the line of identity (b = 1) for the neural network or the Bayesian analysis. The area under the receiver operating characteristic curve was 0.871 for the neural network and 0.894 for the Bayesian analysis (P < .05). There were 23 and 21 false-positive predictions and 18 and six false-negative predictions for the neural network and Bayesian analysis, respectively. CONCLUSION: The Bayesian method was better than the neural network in prediction of probability of malignancy in solitary pulmonary nodules.

Adult↗
PubMed1995

A neural network model for survival data.

Neural networks have received considerable attention recently, mostly by non-statisticians. They are considered by many to be very promising tools for classification and prediction. In this paper we present an approach to modelling censored survival data using the input-output relationship associated with a simple feed-forward neural network as the basis for a non-linear proportional hazards model. This approach can be extended to other models used with censored survival data. The proportional hazards neural network parameters are estimated using the method of maximum likelihood. These maximum likelihood based models can be compared, using readily available techniques such as the likelihood ratio test and the Akaike criterion. The neural network models are illustrated using data on the survival of men with prostatic carcinoma. A method of interpreting the neural network predictions based on the factorial contrasts is presented.

Diethylstilbestrol↗
PubMed1995

Pediatric skeletal age: determination with neural networks.

PURPOSE: To develop a neural network to calculate skeletal age based on measurements taken from digitized hand radiographs. MATERIALS AND METHODS: From a database of 521 hand radiographs obtained in healthy patients, four parameters were calculated from seven linear measurements and were used to train a neural network, with use of the jackknife method, to calculate skeletal age. The results were compared with those of an experienced pediatric radiologist using a standard pediatric skeletal atlas. RESULTS: The mean difference from biologic age for the neural network was -0.261 years +/- 1.82 (standard deviation) and for the radiologist, -0.232 years +/- 1.54; this difference was not significantly different (P = .67, Wilcoxon signed rank test). Skeletal age determined by the neural network was closer to the biologic age than that assigned by the radiologist in 243 of 521 cases (47%). CONCLUSION: A simple neural network may assist radiologists in the assessment of skeletal age.

Adolescent↗