Computing with neural networks.
Explore the source record for details and available documents.
Artificial Intelligence↗
SEARCH · Search PubMed
Results for “Neural Networks, Computer”
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.
PAPNET for analysis of proliferating (MIB-1 positive) cell populations in cervical smears.
In diagnostic tumor pathology, immunohistochemical detection of proliferating cell populations is increasingly used. With the event of microwave-antigen retrieval it has become possible to detect proliferating cells in cervical smears staining positive for MiB-1. We report that with the PAPNET system, using neural network computing, it is possible to collect from the smears the images with epithelial fragments containing positive nuclei. We used this system for quantification of staining results. Cases with carcinoma in situ contained many epithelial fragments having a large number of positive-staining nuclei and with labelling indices of 60 +/- 16. Dysplasias were often completely devoid of cells with positive nuclei. In addition, we could not detect differences between progressive and regressive dysplasias. Automatic prescreening of immunostained smears using PAPNET is useful when it is desired to quantify staining results.
Antibodies, Monoclonal↗
Semiautomatic PAPNET analysis of proliferating (MiB-1-positive) cells in cervical cytology and histology.
With the event of microwave-antigen retrieval it has become possible to detect proliferating cells (staining positive for MiB-1) in cervical smears containing epithelial fragments and in paraffin sections containing cancerous cervical epithelium. The PAPNET system, using neural network computing, is able to collect from the slides epithelial fragments with positive-staining nuclei. The nuclei in epithelial fragments are semiautomatically quantitated using the PAPNET-digitized images. The parameter PPN (proportion-positive nuclei), in which the nuclear area of the positive nuclei is taken into account, prove to be superior to the proliferation index for distinguishing moderate dysplasia from carcinoma in situ. In repair cells, all four quantitative parameters are close to those of moderate dysplasias, indicating that this method is unfit for distinguishing these two entities. However, we show that MiB-1 staining is valuable for the analysis of "false-negative" and "false-positive" smears, and for quantifying proliferation in sections of carcinoma in situ.
Antigens, Nuclear↗
Hippocampus, cortex, and basal ganglia: insights from computational models of complementary learning systems.
We present a framework for understanding how the hippocampus, neocortex, and basal ganglia work together to support cognitive and behavioral function in the mammalian brain. This framework is based on computational tradeoffs that arise in neural network models, where achieving one type of learning function requires very different parameters from those necessary to achieve another form of learning. For example, we dissociate the hippocampus from cortex with respect to general levels of activity, learning rate, and level of overlap between activation patterns. Similarly, the frontal cortex and associated basal ganglia system have important neural specializations not required of the posterior cortex system. Taken together, this overall cognitive architecture, which has been implemented in functioning computational models, provides a rich and often subtle means of explaining a wide range of behavioral and cognitive neuroscience data. Here, we summarize recent results in the domains of recognition memory, contextual fear conditioning, effects of basal ganglia lesions on stimulus-response and place learning, and flexible responding.
Animals↗
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↗
A neural networks-based drug discovery approach and its application for designing aldose reductase inhibitors.
A novel approach that combines neural networks, computer docking and quantum mechanical method is developed to design potent aldose reductase inhibitors (ARIs). Neural networks is employed to determine the quantitative structure-activity relationship (QSAR) among the known ARIs. The physical descriptors of the neural networks, such as electronegativity and molar volume, are evaluated with first-principles quantum mechanical method. Based on the QSAR, new candidates for ARI are predicted, and subsequently screened via computer docking technique. The surviving candidates are further tested via quantum mechanical calculation for their bindings to aldose reductase. We find that the best 49 predicted ARI candidates have better calculated binding energies than those of experimentally known drug candidates.
Aldehyde Reductase↗
Simulation of the 13C nuclear magnetic resonance spectra of trisaccharides using multiple linear regression analysis and neural networks.
Predictive models are developed for the 13C NMR chemical shifts of the carbon atoms comprising the central rings of 46 trisaccharide compounds. Thirty-nine trisaccharides are used as a training set for development of models using regression analysis and computational neural networks, and seven compounds are used as an external prediction set. The descriptors used in the models are developed directly from the molecular structures of the trisaccharides. Three different methods of descriptor selection are compared. The dependence of the models on the geometries of the trisaccharides is explored. The models developed with geometric descriptors are better than those developed without geometric descriptors, although the latter models are still of a comparable quality. Overall, the best model found is a neural network based on descriptors selected by multiple linear regression.
Algorithms↗
A theory for cursive handwriting based on the minimization principle.
We propose a trajectory planning and control theory which provides explanations at the computation, algorithm, representation, and hardware levels for continuous movement such as connected cursive handwriting. The hardware is based on our previously proposed forward-inverse-relaxation neural network. Computationally, the optimization principle is the minimum torque-change criterion. At the representation level, hard constraints satisfied by a trajectory are represented as a set of via-points extracted from handwritten characters. Accordingly, we propose a via-point estimation algorithm that estimates via-points by repeating trajectory formation of a character and via-point extraction from the character. It is shown experimentally that for movements with a single via-point target, the via-point estimation algorithm can assign a point near the actual via-point target. Good quantitative agreement is found between human movement data and the trajectories generated by the proposed model.
Algorithms↗
Molecular descriptors that influence the amount of drugs transfer into human breast milk.
Most drugs are excreted into breast milk to some extent and are bioavailable to the infant. The ability to predict the approximate amount of drug that might be present in milk from the drug structure would be very useful in the clinical setting. The aim of this research was to simplify and upgrade the previously developed model for prediction of the milk to plasma (M/P) concentration ratio, given only the molecular structure of the drug. The set of 123 drug compounds, with experimentally derived M/P values taken from the literature, was used to develop, test and validate a predictive model. Each compound was encoded with 71 calculated molecular structure descriptors, including constitutional descriptors, topological descriptors, molecular connectivity, geometrical descriptors, quantum chemical descriptors, physicochemical descriptors and liquid properties. Genetic algorithm was used to select a subset of the descriptors that best describe the drug transfer into breast milk and artificial neural network (ANN) to correlate selected descriptors with the M/P ratio and develop a QSAR. The averaged literature M/P values were used as the ANN's output and calculated molecular descriptors as the inputs. A nine-descriptor nonlinear computational neural network model has been developed for the estimation of M/P ratio values for a data set of 123 drugs. The model included the percent of oxygen, parachor, density, highest occupied molecular orbital energy (HOMO), topological indices (chiV2, chi2 and chi1) and shape indices (kappa3, kappa2), as the inputs had four hidden neurons and one output neuron. The QSPR that was developed indicates that molecular size (parachor, density) shape (topological shape indices, molecular connectivity indices) and electronic properties (HOMO) are the most important for drug transfer into breast milk. Unlike previously reported models, the QSPR model described here does not require experimentally derived parameters and could potentially provide a useful prediction of M/P ratio of new drugs only from a sketch of their structure and this approach might also be useful for drug information service. Regardless of the model or method used to estimate drug transfer into breast milk, these predictions should only be used to assist in the evaluation of risk, in conjunction with assessment of the infant's response.
Animals↗
Pupillary assessment and computational modeling of the Stroop task in depression.
Depressed individuals frequently display disruptions in selective attention, but the time course and specificity of these difficulties are not well-understood. To better understand the nature of attentional disruptions in depression, 28 healthy adults and 23 unmedicated depressed adults completed a Stroop color-naming task using a long inter-stimulus interval and pupil dilation was recorded as a measure of cognitive load. Both groups took longer to name the color for incongruent than congruent trials. Pupil dilation was also larger for incongruent trials than for congruent trials across groups, which suggested that pupil dilation reflected cognitive load on the task. Though the groups did not differ in the magnitude of Stroop effect in pupil dilation, depressed individuals displayed decreased pupil dilation in the seconds following stimuli relative to controls. Computational neural network modeling further suggested that observed effects were consistent with decreased prefrontal cortex activity, associated with decreased cognitive control.
Adolescent↗
Evaluation of prolonged fetal monitoring with normal and pathologic outcome probabilities determined by artificial neural network.
OBJECTIVE: The purpose of this study was to objectively evaluate prolonged fetal heart rate (FHR) monitoring, which has been difficult to do with conventional cardiotocogram (CTG). METHODS: FHR was analyzed by an artificial neural network computer that calculates probabilities of normal, pathologic and suspicious outcome. Earlier normal and pathologic outcome probabilities (OPs) recorded during 15-min intervals are averaged every 5 min. Initially, two curves (the averaged normal and pathologic OPs) are compared. Furthermore, a single curve traced for each difference between the averaged normal and pathologic OPs and its value are studied. Our FHR probability data are of 9 cases reported in a previous paper on neural network FHR analysis. RESULTS: In the 4 cases of normal neonatal condition, the trends of the averaged curves and the last averaged values were higher for normal OP than for pathologic OP, and the final values of the difference were >0. On the other hand, in the 5 cases of neonatal depression, the trend of the two curves and the final values were lower for normal than for pathologic OP; and the final difference values of averaged probabilities were <0. For prolonged monitoring, the single parameter is more useful than the comparison of the two curves. CONCLUSION: A useful single parameter is obtained for the accurate and objective evaluation of prolonged FHR monitoring. The present method is promising for prospective studies using the combined system of experts and neural computers.
Female↗
Reinforcement learning with internal expectation in the random neural networks for cascaded decisions.
The reinforcement learning scheme proposed in Halici (J. Biosystems 40 (1997) 83) for the random neural network (RNN) (Neural Computation 1 (1989) 502) is based on reward and performs well for stationary environments. However, when the environment is not stationary it suffers from getting stuck to the previously learned action and extinction is not possible. To overcome the problem, the reinforcement scheme is extended in Halici (Eur. J. Oper. Res., 126(2000) 288) by introducing a new weight update rule (E-rule) which takes into consideration the internal expectation of reinforcement. Although the E-rule is proposed for the RNN, it can be used for training learning automata or other intelligent systems based on reinforcement learning. This paper looks into the behavior of the learning scheme with internal expectation for the environments where the reinforcement is obtained after a sequence of cascaded decisions. The simulation results have shown that the RNN learns well and extinction is possible even for the cases with several decision steps and with hundreds of possible decision paths.
Computer Simulation↗
The elusive concept of brain connectivity.
Neurons and neural populations do not function as islands onto themselves. Rather, they interact with other such elements through their afferent and efferent connections in an orchestrated manner so as to enable different sensorimotor and cognitive tasks to be performed. The concept of functional connectivity and the allied notion of effective connectivity were introduced to designate the functional strengths of such interactions. Functional neuroimaging methods, especially PET and fMRI, have been used extensively to evaluate the functional connectivity between different brain regions. After providing a brief historical review of these notions of brain connectivity, I argue that the conceptual formulations of functional and effective connectivity are far from clear. Specifically, the terms functional and effective connectivity are applied to quantities computed on types of functional imaging data (e.g., PET, fMRI, EEG) that vary in spatial, temporal, and other features, using different definitions (even for data of the same modality) and employing different computational algorithms. Until it is understood what each definition means in terms of an underlying neural substrate, comparisons of functional and/or effective connectivity across studies may appear inconsistent and should be performed with great caution.
Brain↗
Steroid binding by antibodies and artificial receptors: exploration of theoretical methods to determine the origins of binding affinities and specificities.
Binding mode calculations for complexes between an artificial paracyclophane receptor and digoxins, cholic acids as well as cortisone steroids show encapsulation of different ring combinations. Docking experiments were performed between the 26-10 antibody and digoxins. Coordination affinity arises from hydrophobic desolvation and van der Waals interactions rather than from hydrogen bonds. The specificity and affinity arises mainly from shape complementarity. Computed binding free energies and Kohonen neural network computations both point to physicochemical and structural similarities of natural antibodies and artificial receptors.
Antibodies↗
Synaptic elimination, neurodevelopment, and the mechanism of hallucinated "voices" in schizophrenia.
OBJECTIVE: After peaking during childhood, synaptic density in the human frontal cortex declines by 30%-40% during adolescence because of progressive elimination of synaptic connections. The characteristic age at onset of schizophrenia--late adolescence and early adulthood--suggests that the disorder could arise from irregularities involving this neurodevelopmental process. METHOD: A computer simulation of a speech perception neural network was developed. Connections within the working memory component of the network were eliminated on the basis of a "Darwinian rule" in order to model loss of synapses. As a comparison, neuronal cell death, also postulated as being linked to both neurodevelopment and schizophrenia, was simulated. The authors determined whether these alterations at low levels could enhance perceptual capacity and at high levels produce spontaneous speech percepts that simulate hallucinated speech or "voices." RESULTS: Eliminating up to 65% of working memory connections improved perceptual ability; beyond that point, network performance declined and speech hallucinations emerged. Simulating excitotoxic neuronal loss at low levels also improved network performance, but in excess it did not produce hallucinations. CONCLUSIONS: The model demonstrates perceptual advantages of selective synaptic elimination as well as selective neuronal loss, suggesting a functional explanation for these aspects of neurodevelopment. The model predicts that psychosis arises from a pathological extension of one of these neurodevelopmental trends, namely, synaptic elimination.
Auditory Perception↗
Receptive field changes after strokelike cortical ablation: a role for activation dynamics.
The reorganization of neural activity that takes place after stroke is of paramount importance in producing functional recovery. Experimental stroke models have suggested that this reorganization may have two phases, but physiology alone cannot fully resolve what causes each phase. Computer modeling suggests that these phases might involve an initial change in dynamics occurring immediately, followed by synaptic plasticity. We combined physiological recording from macaque middle temporal cortex (area MT) with a neural network computer model to examine this first phase of altered cortical function after a small, experimentally induced cortical lesion. Major receptive field (RF) changes seen in the first few days postlesion included both expansion and contraction of receptive fields. Although only expansion could be reproduced in an initial model, addition of inhibitory interneuron loss in a ring around the primary ablation, suggested by immunohistochemical examination, permitted contraction to be replicated as well. We therefore predict that this immunochemical observation reflects an immediate extension of the lesion rather than a late response. Additionally our model successfully predicted a correlation between increased firing rate and RF size. Our model suggests that activation dynamics alone, without anatomic remodeling, can cause the large receptive field changes that allow the rapid behavioral recovery seen after middle temporal lesions.
Animals↗
Differentiation of beats of ventricular and sinus origin using a self-training neural network.
Despite advances in the computerized detection of arrhythmias, arrhythmia recognition by morphological waveform analysis still poses a difficult problem. Artificial neural networks, computer algorithms that are self-trained by an analog of biological synaptic modification to perform pattern recognition, hold great promise for the differentiation of various cardiac rhythms. The goal of this study was to differentiate beats of sinus and ventricular origin on a global basis and on a patient-specific basis by the use of artificial neural network analysis. Neural networks were trained to recognize digitized intracardiac electrograms (9 patients) and surface electrocardiograms (11 patients) obtained during sinus rhythm and ventricular tachycardia. After training, sinus rhythm or ventricular tachycardia beats were input into the neural network, and classified as to their origin. By the use of modified receiver operating characteristic curve plots, it was possible to differentiate with high sensitivity and specificity between beats of sinus origin and ventricular origin in all patients. The addition of high amounts of noise to the beats did not markedly degrade the performance of the surface ECG neural networks, and still allowed high sensitivity in differentiating beats of sinus origin from beats of ventricular origin, especially when noise was added to the training set. Neural networks provided sensitive and specific detection of cardiac electrical activity during sinus rhythm and ventricular tachycardia, and may play an important role in allowing development of improved arrhythmia recognition and management systems.
Aged↗
Use of artificial neural networks in a QSAR study of anti-HIV activity for a large group of HEPT derivatives.
Anti-HIV activity for a set of 107 inhibitors of the HIV-1 reverse transcriptase, derivatives of 1-[2-hydroxyethoxy)methyl]-6-(phenylthio)thymine (HEPT), was modeled with the aid of chemometric techniques. The activity of these compounds was estimated by means of multiple linear regression (MLR) and artificial neural network (ANN) techniques and compared with the previous works. The results obtained using the MLR method indicate that the anti-HIV activity of the HEPT derivatives depends on the reverse of standard shadow area on the YZ plane and the ratio of the partial charges of the most positive atom to the most negative atom of the molecule. The best computational neural network model was a fully-connected, feed-forward method with a 6-6-1 architecture. The mean-square error for the prediction set using this network was 0.372 compared with 0.780 obtained using the MLR technique. Comparison of the quality of the ANN of this work with different MLR models shows that ANN has a better predictive power.
Anti-HIV Agents↗