Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Ensemble methods”

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 271 records · Page 15Linked to original sources
PubMed2003

Training in cortical control of neuroprosthetic devices improves signal extraction from small neuronal ensembles.

We have recently developed a closed-loop environment in which we can test the ability of primates to control the motion of a virtual device using ensembles of simultaneously recorded neurons /29/. Here we use a maximum likelihood method to assess the information about task performance contained in the neuronal ensemble. We trained two animals to control the motion of a computer cursor in three dimensions. Initially the animals controlled cursor motion using arm movements, but eventually they learned to drive the cursor directly from cortical activity. Using a population vector (PV) based upon the relation between cortical activity and arm motion, the animals were able to control the cursor directly from the brain in a closed-loop environment, but with difficulty. We added a supervised learning method that modified the parameters of the PV according to task performance (adaptive PV), and found that animals were able to exert much finer control over the cursor motion from brain signals. Here we describe a maximum likelihood method (ML) to assess the information about target contained in neuronal ensemble activity. Using this method, we compared the information about target contained in the ensemble during arm control, during brain control early in the adaptive PV, and during brain control after the adaptive PV had settled and the animal could drive the cursor reliably and with fine gradations. During the arm-control task, the ML was able to determine the target of the movement in as few as 10% of the trials, and as many as 75% of the trials, with an average of 65%. This average dropped when the animals used a population vector to control motion of the cursor. On average we could determine the target in around 35% of the trials. This low percentage was also reflected in poor control of the cursor, so that the animal was unable to reach the target in a large percentage of trials. Supervised adjustment of the population vector parameters produced new weighting coefficients and directional tuning parameters for many neurons. This produced a much better performance of the brain-controlled cursor motion. It was also reflected in the maximum likelihood measure of cell activity, producing the correct target based only on neuronal activity in over 80% of the trials on average. The changes in maximum likelihood estimates of target location based on ensemble firing show that an animal's ability to regulate the motion of a cortically controlled device is not crucially dependent on the experimenter's ability to estimate intention from neuronal activity.

Action Potentials↗
PubMed1999

Structural transitions in the protein L denatured state ensemble.

We use a broad array of biophysical methods to probe the extent of structure and time scale of structural transitions in the protein L denatured state ensemble. Measurement of amide proton exchange protection during the first several milliseconds following initiation of refolding in 0.4 M sodium sulfate revealed weak protection in the first beta-hairpin and helix. A tryptophan residue was introduced into the first beta-hairpin to probe the extent of structure formation in this part of the protein; the intrinsic fluorescence of this tryptophan was found to deviate from that expected given its local sequence context in 2-3 M guanidine, suggesting some partial ordering of this region in the unfolded state ensemble. To further probe this partial ordering, dansyl groups were introduced via cysteine residues at three sites in the protein. It was found that fluorescence energy transfer from the introduced tryptophan to the dansyl groups decreased dramatically upon unfolding. Stopped-flow fluorescence studies showed that the recovery of dansyl fluorescence upon refolding occurred on a submillisecond time scale. To probe the interactions responsible for the residual structure observed in the denatured state ensemble, the conformation of a peptide corresponding to the first beta-hairpin and helix of protein L was studied using circular dichroism spectroscopy and compared to that of full-length protein L and previously characterized peptides corresponding to the isolated helix and second beta-hairpin.

Amino Acid Sequence↗
PubMed2003

Detection of mutual phase synchronization in multivariate signals and application to phase ensembles and chaotic data.

This work presents a method for the detection of mutual phase synchronization in nonstationary time series. We show how the application of a cluster algorithm that considers spatiotemporal structures of data follows from the general condition of phase-synchronized data. In view of the topology of phasic data, we reformulate the K-means cluster algorithm on a flat torus and apply a segmentation index derived in an earlier work [A. Hutt and H. Riedel, Physica D 177, 203 (2003)]. This index is extended by means of averaging in order to reflect phase synchronization in ensembles of multivariate time series. The method is illustrated using simulated multivariate phase dynamics and arrays of chaotic systems, in which temporal segments of phase-synchronized states are registered. A comparison with results from an existing bivariate synchronization index reveals major advantages of our method.

Journal Article↗
PubMed1998

Design of Protein Cores by Screening Combinatorial Sequence Library.

We have developed a new method, heterogeneous self-consistent ensemble optimization (hetero-SCEO), to select appropriate hydrophobic cores of proteins. It has been tested with five kinds of proteins: lambda-repressor, phage 434 CRO protein, interleukin-4, thioredoxin and ubiquitin. The results show that the method can be used for the de novo desigh of the hydrophobic cores of proteins.

Journal Article↗
PubMed2005

Firing rate modulation: a simple statistical view of memory trace reactivation.

Memory trace reactivation in hippocampal ensembles during sleep has been suggested as a coordinating mechanism for consolidation of new memories. Here we propose a simple statistical scheme allowing analysis of the reactivation of firing rate modulations, with a well-defined null hypothesis. This method allowed reliable detection of ensemble reactivation across three experimental settings. Reactivation of firing rate modulations mirrors several properties of commonly studied reactivation measures: it is stronger during hippocampal sharp waves, and decays over a period of 10-20 min. Moreover, in some conditions, firing rate reactivation covaries with reactivation of cell pair cross-correlations, suggesting the two phenomena reflect similar processes. We propose an attractor network model, with pre-wired attractors, in which experience selects and primes some attractors. Priming occurs by either experience dependent synaptic plasticity or changes in neuronal excitability. Primed attractors are more likely to activate in the following sleep, inducing reactivation of both rates and cross-correlations.

Animals↗
PubMed1995

Improving the ensemble average of visual evoked potentials. I. Theory of correction by spectral phase difference methods.

The considered causes of variability of visual evoked potentials (VEPs) are additive noise and time variability. The time variability can be modelled in three different ways: 1. time invariant VEP waveform and delay-time "jitter"; 2. time variant VEP waveform, due to 2a. jitter of individual components or 2b. constant VEP waveform and "wow", i.e. expansion or compression of the time scale. The wow model has not been considered so far in the literature. Correction of jitter and/or wow is based on a frequency domain method: the "spectral phase difference" method (SPD). When jitter is present the SPD method is applied in the frequency domain on the single responses, or on a subensemble of a small number of responses. For the wow a new method is proposed where the time scale is logarithmically compressed prior to the calculation of the spectrum. It is shown that the SPD method can also be applied then to estimate the compression factor yielding the wow. The single responses, or the averaged sub-ensembles, can subsequently be corrected by this factor and finally be averaged. Improvement of the precision and accuracy of estimates of both the peak latencies and the peak amplitudes, as well as a comparison between the results of the jitter and wow models, have been reported in a companion paper.

Computer Simulation↗
PubMed2000

Speckle tracking for multi-dimensional flow estimation.

Speckle tracking methods overcome the major limitations of current Doppler methods for flow imaging and quantification: angle dependence and aliasing. In this paper, we review the development of speckle tracking, with particular attention to the advantages and limitations of two-dimensional algorithms that use a single transducer aperture. Ensemble tracking, a recent speckle tracking method based upon parallel receive processing, is described. Experimental results with ensemble tracking indicate the ability to measure laminar flow in a phantom at a beam-vessel angle of 60 degrees, which had not been possible with previous 2D speckle tracking methods. Finally, important areas for future research in speckle tracking are briefly summarized.

Algorithms↗
PubMed1994

Parametric and ensemble sequence alignment algorithms.

Recently algorithms for parametric alignment (Waterman et al., 1992, Natl Acad. Sci. USA 89, 6090-6093; Gusfield et al., 1992, Proceedings of the Third Annual ACM-SIAM Discrete Algorithms) find optimal scores for all penalty parameters, both for global and local sequence alignment. This paper reviews those techniques. Then in the main part of this paper dynamic programming methods are used to compute ensemble alignment, finding all alignment scores for all parameters. Both global and local ensemble alignments are studied, and parametric alignment is used to compute near optimal ensemble alignments.

Algorithms↗
PubMed1999

Engineering Novel Functional Proteins Grafting Active Sites into Natural Scaffolds.

Engineering novel small functional proteins by grafting active sites into small but stable proteins is an efficient protein design method. Combining heterogeneous self-consistent ensemble optimization (hetero-SCEO) with 3D-motif search tool, we developed a system to accomplish such method. It is tested by transferring zinc-binding site of carbonic anhydrase form B to charybdotoxin and its efficiency is demonstrated.

Journal Article↗
PubMed2005

Phase behavior of Ising mixtures.

We present phase diagrams that were calculated both in mean-field theory and via Monte Carlo (MC) simulations for binary mixtures of a ferromagnetic Ising fluid and a nonmagnetic fluid (Ising mixtures) in the absence of an external field. We look at both the simple ideal Ising mixture, consisting of an ideal Ising fluid and a hard-sphere fluid, as well as at the general case with one component being a nonideal Ising fluid and the other a van der Waals fluid. It is shown that the mean-field phase diagram of the ideal Ising mixture in the limit of infinite pressure is identical to that of the Blume-Capel model for 3He-4He mixtures. The MC phase diagrams were obtained using the Gibbs ensemble, the cumulant intersection technique, and the multi-histogram re-weighting method, adapted to the semi-grand ensemble. The results are qualitatively compared with mean-field theory, and both types of tri-critical lines occurring there are verified in the computer simulations.

Journal Article↗
PubMed2006

Ensemble-based discriminant learning with boosting for face recognition.

In this paper, we propose a novel ensemble-based approach to boost performance of traditional Linear Discriminant Analysis (LDA)-based methods used in face recognition. The ensemble-based approach is based on the recently emerged technique known as "boosting". However, it is generally believed that boosting-like learning rules are not suited to a strong and stable learner such as LDA. To break the limitation, a novel weakness analysis theory is developed here. The theory attempts to boost a strong learner by increasing the diversity between the classifiers created by the learner, at the expense of decreasing their margins, so as to achieve a tradeoff suggested by recent boosting studies for a low generalization error. In addition, a novel distribution accounting for the pairwise class discriminant information is introduced for effective interaction between the booster and the LDA-based learner. The integration of all these methodologies proposed here leads to the novel ensemble-based discriminant learning approach, capable of taking advantage of both the boosting and LDA techniques. Promising experimental results obtained on various difficult face recognition scenarios demonstrate the effectiveness of the proposed approach. We believe that this work is especially beneficial in extending the boosting framework to accommodate general (strong/weak) learners.

Algorithms↗
PubMedDate not supplied

Using ensembles to classify compounds for drug discovery.

This paper introduces Signal, a novel method for classifying activity against a small molecule drug target. Signal creates an ensemble, or collection, of meaningful descriptors chosen from a much larger property space. The method works with a variety of descriptor types, including fingerprints that represent four-point pharmacophores or shape descriptors. It also exploits information from both active and inactive compounds and generates predictive models suitable for high throughput screening data analysis. Given the fingerprints and activity data for a set of compounds, Signal is a two step process. The first step is to Evaluate the Descriptors: for each descriptor in the fingerprint, quantify and rank the correlation between the activity of the compounds and the presence of that descriptor. The second step is to Create an Ensemble Model: use the high ranking descriptors to create a model of activity against the biological target. For the first step, two possible ranking strategies were investigated: mutual information and chi-square. For the second step, two types of ensemble models were investigated: high ranking and a novel method called high ranking set cover. Of the four possible pairings, the combination of chi-square and high ranking set cover performed the best on a Thrombin data set.

Algorithms↗
PubMed2005

Interpreting dynamically-averaged scalar couplings in proteins.

The experimental determination of scalar three-bond coupling constants represents a powerful method to probe both the structure and dynamics of proteins. The detailed structural interpretation of such coupling constants is usually based on Karplus relationships, which allow the measured couplings to be related to the torsion angles of the molecules. As the measured couplings are sensitive to thermal fluctuations, the parameters in the Karplus relationships are better derived from ensembles representing the distributions of dihedral angles present in solution, rather than from single conformations. We present a method to derive such parameters that uses ensembles of conformations determined through dynamic-ensemble refinement--a method that provides structural ensembles that simultaneously represent both the structure and the associated dynamics of a protein.

Models, Molecular↗
PubMed2003

Ensemble machine learning on gene expression data for cancer classification.

Whole genome RNA expression studies permit systematic approaches to understanding the correlation between gene expression profiles to disease states or different developmental stages of a cell. Microarray analysis provides quantitative information about the complete transcription profile of cells that facilitate drug and therapeutics development, disease diagnosis, and understanding in the basic cell biology. One of the challenges in microarray analysis, especially in cancerous gene expression profiles, is to identify genes or groups of genes that are highly expressed in tumour cells but not in normal cells and vice versa. Previously, we have shown that ensemble machine learning consistently performs well in classifying biological data. In this paper, we focus on three different supervised machine learning techniques in cancer classification, namely C4.5 decision tree, and bagged and boosted decision trees. We have performed classification tasks on seven publicly available cancerous microarray data and compared the classification/prediction performance of these methods. We have observed that ensemble learning (bagged and boosted decision trees) often performs better than single decision trees in this classification task.

Algorithms↗
PubMed2006

The nature of the calculation of the pressure in molecular simulations of continuous models from volume perturbations.

We consider some fundamental aspects of the calculation of the pressure from simulations by performing volume perturbations. The method, initially proposed for hard-core potentials by Eppenga and Frenkel [Mol. Phys.52, 1303 (1984)] and then extended to continuous potentials by Harismiadis et al. [J. Chem. Phys. 105, 8469 (1996)], is based on the numerical estimate of the change in Helmholtz free energy associated with the perturbation which, in turn, can be expressed as an ensemble average of the corresponding Boltzmann factor. The approach can be easily generalized to the calculation of components of the pressure tensor and also to ensembles other than the canonical ensemble. The accuracy of the method is assessed by comparing simulation results obtained from the volume-perturbation route with those obtained from the usual virial expression for several prototype fluid models. Monte Carlo simulation data are reported for bulk fluids and for inhomogeneous systems containing a vapor-liquid interface.

Journal Article↗
PubMed1998

Alterations in information transmission in ensembles of primary muscle spindle afferents after muscle fatigue in heteronymous muscle.

This study showed that fatigue of the ipsilateral medial gastrocnemius muscle caused a clear-cut reduction in the ability of ensembles of primary muscle spindle afferents from the lateral gastrocnemius muscle to discriminate between muscle stretches of varying amplitude. The results were probably caused by reflex-mediated effects from chemosensitive group III and IV afferents onto the gamma-motoneurons projecting to lateral gastrocnemius muscle spindles. The experiments were conducted on seven cats anaesthetized with alpha-chloralose and a total of 41 primary muscle spindle afferents from the lateral gastrocnemius were registered. Afferents were simultaneously recorded in ensembles of three to 10 afferents. A method based on principal component analysis and algorithms for quantification of stimulus discrimination in ensembles of muscle afferents was used prior to, immediately following and five or more minutes after muscle fatigue had been induced to the ipsilateral medial gastrocnemius muscle. It is well established that the primary muscle spindle afferents play an important role in proprioception and kinaesthesia. Therefore the decrease in the accuracy of the information transmitted by ensembles of primary muscle spindle afferents caused by fatigue in an ipsilateral muscle implies concomitant effects on proprioception and kinaesthesia.

Animals↗
PubMed2001

Isobaric molecular dynamics simulations of hard sphere systems.

We describe an implementation of the Andersen algorithm for simulating the molecular dynamics in the isobaric isoenthalpic (NPH) ensemble for the hard sphere potential. The work is based on the adaptation of the Andersen algorithm to hard spheres by de Smedt et al. For a hard sphere system in the NPH ensemble, the particle velocities are not constant between collisions and we describe an efficient method for handling this part of the dynamics. The method is extended to give an NPT ensemble simulation of hard sphere systems by applying an ad hoc rescaling of the velocities. The accuracy of the algorithms is tested by comparison with traditional NVE simulation results for the structural, thermodynamic, and transport properties.

Journal Article↗