Search PubMed⌕ Search

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.

At least 307 records · Page 17Linked to original sources
PubMed2000

Quantitative structure-property relationships for the prediction of vapor pressures of organic compounds from molecular structures

A quantitative structure-property relationship (QSPR) is developed to relate the molecular structures of 420 diverse organic compounds to their vapor pressures at 25 degrees C expressed as log(vp), where vp is in pascals. The log(vp) values range over 8 orders of magnitude from -1.34 to 6.68 log units. The compounds are encoded with topological, electronic, geometrical, and hybrid descriptors. Statistical and computational neural network (CNN) models are built using subsets of the descriptors chosen by simulated annealing and genetic algorithm feature selection routines. An 8-descriptor CNN model, which contains only topological descriptors, is presented which has a root-mean-square (rms) error of 0.37 log unit for a 65-member external prediction set. A 10-descriptor CNN model containing a larger selection of descriptor types gives an improved rms error of 0.33 log unit for the external prediction set.

Journal Article↗
PubMed1998

Frequency control in synchronized networks of inhibitory neurons.

We analyze the control of frequency for a synchronized inhibitory neuronal network. The analysis is done for a reduced membrane model with a biophysically based synaptic influence. We argue that such a reduced model can quantitatively capture the frequency behavior of a larger class of neuronal models. We show that in different parameter regimes, the network frequency depends in different ways on the intrinsic and synaptic time constants. Only in one portion of the parameter space, called phasic, is the network period proportional to the synaptic decay time. These results are discussed in connection with previous work of the authors, which showed that for mildly heterogeneous networks, the synchrony breaks down, but coherence is preserved much more for systems in the phasic regime than in the other regimes. These results imply that for mildly heterogeneous networks, the existence of a coherent rhythm implies a linear dependence of the network period on synaptic decay time and a much weaker dependence on the drive to the cells. We give experimental evidence for this conclusion.

Action Potentials↗
PubMedDate not supplied

Turning on and off with excitation: the role of spike-timing asynchrony and synchrony in sustained neural activity.

Delay-related sustained activity in the prefrontal cortex of primates, a neurological analogue of working memory, has been proposed to arise from synaptic interactions in local cortical circuits. The implication is that memories are coded by spatially localized foci of sustained activity. We investigate the mechanisms by which sustained foci are initiated, maintained, and extinguished by excitation in networks of Hodgkin-Huxley neurons coupled with biophysical spatially structured synaptic connections. For networks with a balance between excitation and inhibition, a localized transient stimulus robustly initiates a localized focus of activity. The activity is then maintained by recurrent excitatory AMPA-like synapses. We find that to maintain the focus, the firing must be asynchronous. Consequently, inducing transient synchrony through an excitatory stimulus extinguishes the sustained activity. Such a monosynaptic excitatory turn-off mechanism is compatible with the working memory being wiped clean by an efferent copy of the motor command. The activity that codes working memories may be structured so that the motor command is both the read-out and a direct clearing signal. We show examples of data that is compatible with our theory.

Action Potentials↗
PubMedDate not supplied

Unsupervised learning and adaptation in a model of adult neurogenesis.

Adult neurogenesis has long been documented in the vertebrate brain and recently even in humans. Although it has been conjectured for many years that its functional role is related to the renewing of memories, no clear mechanism as to how this can be achieved has been proposed. Using the mammalian olfactory bulb as a paradigm, we present a scheme in which incorporation of new neurons proceeds at a constant rate, while their survival is activity-dependent and thus contingent on new neurons establishing suitable connections. We show that a simple mathematical model following these rules organizes its activity so as to maximize the difference between its responses and can adapt to changing environmental conditions in unsupervised fashion, in agreement with current neurophysiological data.

Action Potentials↗
PubMedDate not supplied

Supervised and unsupervised learning with two sites of synaptic integration.

Many learning rules for neural networks derive from abstract objective functions. The weights in those networks are typically optimized utilizing gradient ascent on the objective function. In those networks each neuron needs to store two variables. One variable, called activity, contains the bottom-up sensory-fugal information involved in the core signal processing. The other variable typically describes the derivative of the objective function with respect to the cell's activity and is exclusively used for learning. This variable allows the objective function's derivative to be calculated with respect to each weight and thus the weight update. Although this approach is widely used, the mapping of such two variables onto physiology is unclear, and these learning algorithms are often considered biologically unrealistic. However, recent research on the properties of cortical pyramidal neurons shows that these cells have at least two sites of synaptic integration, the basal and the apical dendrite, and are thus appropriately described by at least two variables. Here we discuss whether these results could constitute a physiological basis for the described abstract learning rules. As examples we demonstrate an implementation of the backpropagation of error algorithm and a specific self-supervised learning algorithm using these principles. Thus, compared to standard, one-integration-site neurons, it is possible to incorporate interesting properties in neural networks that are inspired by physiology with a modest increase of complexity.

Action Potentials↗
PubMedDate not supplied

A model-based approach for the analysis of neuronal information transmission in multi-input and -output systems.

We present a new method to characterize multi-input and output neuronal systems using information theory. To obtain a lower bound of transinformation we take three steps: (1) Estimation of the deterministic response to isolate components carrying stimulus information. The deviation of the original response from the deterministic estimate is defined as noise. (2) Coordinate transformation using PCA yields an uncorrelated representation. (3) Partial transinformation values are calculated independently either by Shannon's formula assuming normality or based on density estimation for arbitrary distributions. We investigate the performance of the algorithms using simulated data and discuss suitable parameter settings. The approach allows to evaluate the degree to which stimulus features are encoded. Its potential is illustrated by analyses of neuronal activity in cat primary visual cortex evoked by electrical retina stimulation.

Action Potentials↗
PubMed2003

Perceptual binding of sensory events: the inclusive characteristics model.

A conceptual model of a perceptual system is proposed, in which each neural level forms characteristics inclusive of the data held in the underlying level. As a result, the stimulus field can be expressed as a hierarchically ordered set of overlying sensory characteristics: from sensory features to higher inclusive characteristics binding sensory data to form whole images and scenes. Specific patterns of electrical activity reflecting inclusive characteristics are transmitted via reverse projections from the upper neural levels to the lower. This forms a downward excitation transmission cascade, stimulating those neural structures whose signals correspond to the higher inclusive characteristics of the given perceptual act. It is demonstrated that these mechanisms are in good agreement with experimental data obtained from psychological and neurophysiological studies and may support the binding of sensory events at all perceptual levels.

Brain↗
PubMedDate not supplied

Optimal stimulus coding by neural populations using rate codes.

We create a framework based on Fisher information for determining the most effective population coding scheme for representing a continuous-valued stimulus attribute over its entire range. Using this scheme, we derive optimal single- and multi-neuron rate codes for homogeneous populations using several statistical models frequently used to describe neural data. We show that each neuron's discharge rate should increase quadratically with the stimulus and that statistically independent neural outputs provides optimal coding. Only cooperative populations can achieve this condition in an informationally effective way.

Action Potentials↗
PubMedDate not supplied

Model of the early development of thalamo-cortical connections and area patterning via signaling molecules.

The mammalian cortex is divided into architectonic and functionally distinct areas. There is growing experimental evidence that their emergence and development is controlled by both epigenetic and genetic factors. The latter were recently implicated as dominating the early cortical area specification. In this paper, we present a theoretical model that explicitly considers the genetic factors and that is able to explain several sets of experiments on cortical area regulation involving transcription factors Emx2 and Pax6, and fibroblast growth factor FGF8. The model consists of the dynamics of thalamo-cortical connections modulated by signaling molecules that are regulated genetically, and by axonal competition for neocortical space. The model can make predictions and provides a basic mathematical framework for the early development of the thalamo-cortical connections and area patterning that can be further refined as more experimental facts become known.

Animals↗
PubMed1997

Effect of relevance shifts in category acquisition: a test of neural networks.

The effect of changing the validity of stimulus dimensions in the course of category learning was investigated. Contrary to previous experiments on interdimensional relevance shifts, a family resemblance stimulus structure was used that allowed for the direct comparison of performance on items learned before and after the relevance shift. In both experiments, the test performance was dominated by information that was acquired after the relevance shift. The results indicate that the acquired knowledge of the 2 learning phases was not integrated, but knowledge learned before the shift was partly replaced by information acquired after the shift. Simulation of the results by the independent cue model, the configural cue model, and ALCOVE (attention learning covering map) demonstrates that these models must be expanded by a mechanism that inhibits the integration of information acquired before and after the relevance shift.

Adult↗
PubMed2001

Conjunctive representations in learning and memory: principles of cortical and hippocampal function.

The authors present a theoretical framework for understanding the roles of the hippocampus and neocortex in learning and memory. This framework incorporates a theme found in many theories of hippocampal function: that the hippocampus is responsible for developing conjunctive representations binding together stimulus elements into a unitary representation that can later be recalled from partial input cues. This idea is contradicted by the fact that hippocampally lesioned rats can learn nonlinear discrimination problems that require conjunctive representations. The authors' framework accommodates this finding by establishing a principled division of labor, where the cortex is responsible for slow learning that integrates over multiple experiences to extract generalities whereas the hippocampus performs rapid learning of the arbitrary contents of individual experiences. This framework suggests that tasks involving rapid, incidental conjunctive learning are better tests of hippocampal function. The authors implement this framework in a computational neural network model and show that it can account for a wide range of data in animal learning.

Animals↗
PubMed2004

Functional frontoparietal connectivity during short-term memory as revealed by high-resolution EEG coherence analysis.

In this electroencephalographic study, the authors modeled the functional connectivity between frontal and parietal areas during short-term memory (STM) processes by spectral coherence analysis and the directed transfer function, that is, for the estimation of coherence "direction." A no-STM task was used as a reference. STM was characterized by an increased frontoparietal electroencephalograph coherence at high frequencies (beta and gamma, 14-45 Hz). In the control task, parietal-to-frontal flow prevailed at those frequencies. However, the STM task showed a bidirectional frontoparietal flow at the gamma band. In conclusion, frontoparietal connectivity would optimize "representational" memory during STM. In this context, the frontal areas would increase their influence on parietal areas for memory retention.

Adult↗
PubMed2003

Synchrony-dependent propagation of firing rate in iteratively constructed networks in vitro.

The precise role of synchronous neuronal firing in signal encoding remains unclear. To examine what kinds of signals can be carried by synchrony, I reproduced a multilayer feedforward network of neurons in an in vitro slice preparation of rat cortex using an iterative procedure. When constant and time-varying frequency signals were delivered to the network, the firing of neurons in successive layers became progressively more synchronous. Notably, synchrony in the in vitro network developed even with uncorrelated input, persisted under a wide range of physiological conditions and was crucial for the stable propagation of rate signals. The firing rate was represented by a classical rate code in the initial layers, but switched to a synchrony-based code in the deeper layers.

Action Potentials↗
PubMed2004

Off-line replay maintains declarative memories in a model of hippocampal-neocortical interactions.

During sleep, neural activity in the hippocampus and neocortex seems to recapitulate aspects of its earlier, awake form. This replay may be a substrate for the consolidation of long-term declarative memories, whereby they become independent of the hippocampus and are stored in neocortex. In contrast to storage, other crucial facets of competent long-term memory, such as maintenance of access to stored traces and preservation of their correct interpretation, have received little attention. We investigate long-term episodic and semantic memory in a theoretical model of neocortical-hippocampal interaction. We find that, in the absence of regular hippocampal reactivation, even supposedly consolidated episodic memories are fragile in the face of cortical semantic plasticity. Replay allows access to episodes stored in the hippocampus to be maintained, by keeping them in appropriate register with changing neocortical representations. Hippocampal storage and replay also has a constructive role in the recall of structured, semantic information.

Brain Injuries↗
PubMed2005

Matching storage and recall: hippocampal spike timing-dependent plasticity and phase response curves.

Hippocampal area CA3 is widely considered to function as an autoassociative memory. However, it is insufficiently understood how it does so. In particular, the extensive experimental evidence for the importance of carefully regulated spiking times poses the question as to how spike timing-based dynamics may support memory functions. Here, we develop a normative theory of autoassociative memory encompassing such network dynamics. Our theory specifies the way that the synaptic plasticity rule of a memory constrains the form of neuronal interactions that will retrieve memories optimally. If memories are stored by spike timing-dependent plasticity, neuronal interactions should be formalized in terms of a phase response curve, indicating the effect of presynaptic spikes on the timing of postsynaptic spikes. We show through simulation that such memories are competent analog autoassociators and demonstrate directly that the attributes of phase response curves of CA3 pyramidal cells recorded in vitro qualitatively conform with the theory.

Action Potentials↗