Independent Ensembles Research Articles

Features that matter: Evolutionary signatures can predict viral transmission routes.

Routes of virus transmission between hosts are key to understanding viral epidemiology. Different routes have large effects on viral ecology, and likelihood and rate of transmission; for example, respiratory and vector-borne viruses together encompass the majority of rapid outbreaks and high-consequence animal and plant epidemics. However, determining the specific transmission route(s) can take months to years, delaying mitigation efforts. Here, we identify the viral features and evolutionary signatures which are predictive of viral transmission routes and use them to predict potential routes for fully-sequenced viruses in silico and rapidly, for both viruses with no observed routes, as well as viruses with missing routes. This was achieved by compiling a dataset of 24,953 virus-host associations with 81 defined transmission routes, constructing a hierarchy of virus transmission encompassing those routes and 42 higher-order modes, and engineering 446 predictive features from three complementary perspectives. We integrated those data and features to train 98 independent ensembles of LightGBM classifiers. We found that all features contributed to the prediction for at least one of the routes and/or modes of transmission, demonstrating the utility of our broad multi-perspective approach. Our framework achieved ROC-AUC = 0.991, and F1-score = 0.855 across all included transmission routes and modes, and was able to achieve high levels of predictive performance for high-consequence respiratory (ROC-AUC = 0.990, and F1-score = 0.864) and vector-borne transmission (ROC-AUC = 0.997, and F1-score = 0.921). Our framework ranks the viral features in order of their contribution to prediction, per transmission route, and hence identifies the genomic evolutionary signatures associated with each route. Together with the more matured field of viral host-range prediction, our predictive framework could: provide early insights into the potential for, and pattern of viral spread; facilitate rapid response with appropriate measures; and significantly triage the time-consuming investigations to confirm the likely routes of transmission.

Visual working memory models of delayed estimation do not generalize to whole-report tasks.

Whole-report working memory tasks provide a measure of recall for all stimuli in a trial and afford single-trial analyses that are not possible with single-report delayed estimation tasks. However, most whole-report studies assume that trial stimuli are encoded and reported independently, and they do not consider the relationships between stimuli presented and reported within the same trial. Here, we present the results of two independently conducted whole-report experiments. The first dataset was recorded by Adam, Vogel, and Awh (2017) and required participants to report color and orientation stimuli using a continuous response wheel. We recorded the second dataset, which required participants to report color stimuli using a set of discrete buttons. We found that participants often group their reports by color similarity, contradicting the assumption of independence implicit in most encoding models of working memory. Next, we showed that this behavior was consistent across participants and experiments when reporting color but not orientation, two circular variables often assumed to be equivalent.Finally, we implemented an alternative to independent encoding where stimuli are encoded as a hierarchical Bayesian ensemble and found that this model predicts biases that are not present in either dataset. Our results suggest that assumptions made by both independent and hierarchical ensemble encoding models-which were developed in the context of single-report delayed estimation tasks-do not hold for the whole-report task. This failure to generalize highlights the need to consider variations in task structure when inferring fundamental principles of visual working memory.

Sensing Coherent Nuclear Spin Dynamics with an Ensemble of Paramagnetic Nitrogen Spins

The unpolarized spin environment surrounding a central spin qubit is typically considered as an incoherent source of dephasing, however, precise characterization and control of the spin bath can yield a resource for storing and sensing with quantum states. In this work, we use nitrogen-vacancy (NV) centers in diamond to measure the coherence of optically dark paramagnetic nitrogen defects (P1 centers) and detect coherent interactions between the P1 centers and a local bath of 13C nuclear spins. The dipolar coupling between the P1 centers and 13C nuclear spins is identified by signature periodic collapses and revivals in the P1 spin coherence signal. We then demonstrate, using a range of dynamical decoupling protocols, that the probing NV centers and the P1 spins are coupled to independent ensembles of 13C nuclear spins. Our work illustrates how the optically dark P1 spins can be used to extract information from their local environment and offers new insight into the interactions within a many-body system. Published by the American Physical Society 2024

Cochlear Implant Artifacts Removal in EEG-Based Objective Auditory Rehabilitation Assessment.

Cochlear implant (CI) is a neural prosthesis that can restore hearing for patients with severe to profound hearing loss. Observed variability in auditory rehabilitation outcomes following cochlear implantation may be due to cerebral reorganization. Electroencephalography (EEG), favored for its CI compatibility and non-invasiveness, has become a staple in clinical objective assessments of cerebral plasticity post-implantation. However, the electrical activity of CI distorts neural responses, and EEG susceptibility to these artifacts presents significant challenges in obtaining reliable neural responses. Despite the use of various artifact removal techniques in previous studies, the automatic identification and reduction of CI artifacts while minimizing information loss or damage remains a pressing issue in objectively assessing advanced auditory functions in CI recipients. To address this problem, we propose an approach that combines machine learning algorithms-specifically, Support Vector Machines (SVM)-along with Independent Component Analysis (ICA) and Ensemble Empirical Mode Decomposition (EEMD) to automatically detect and minimize electrical artifacts in EEG data. The innovation of this research is the automatic detection of CI artifacts using the temporal properties of EEG signals. By applying EEMD and ICA, we can process and remove the identified CI artifacts from the affected EEG channels, yielding a refined signal. Comparative analysis in the temporal, frequency, and spatial domains suggests that the corrected EEG recordings of CI recipients closely align with those of peers with normal hearing, signifying the restoration of reliable neural responses across the entire scalp while eliminating CI artifacts.

Emulating Present and Future Simulations of Melt Rates at the Base of Antarctic Ice Shelves With Neural Networks

AbstractMelt rates at the base of Antarctic ice shelves are needed to drive projections of the Antarctic ice sheet mass loss. Current basal melt parameterizations struggle to link open ocean properties to ice‐shelf basal melt rates for the range of current sub‐shelf cavity geometries around Antarctica. We present a proof of concept exploring the potential of simple deep learning techniques to parameterize basal melt. We train a simple feedforward neural network, or multilayer perceptron, acting on each grid cell separately, to emulate the behavior of circum‐Antarctic cavity‐resolving ocean simulations. We find that this kind of emulator produces reasonable basal melt rates for our training ensemble, at least as close as or closer to the reference than traditional parameterizations. On an independent ensemble of simulations that was produced with the same ocean model but with different model parameters, cavity geometries and forcing, the neural network yields similar results to traditional parameterizations on present conditions. In much warmer conditions, both traditional parameterizations and neural network struggle, but the neural network tends to produce basal melt rates closer to the reference than a majority of traditional parameterizations. While this shows that such a neural network is at least as suitable for century‐scale Antarctic ice‐sheet projections as traditional parameterizations, it also highlights that tuning any parameterization on present‐like conditions can introduce biases and should be used with care. Nevertheless, this proof of concept is promising and provides a basis for further development of a deep learning basal melt parameterization.

Bond stress estimation of profiled steel-concrete in steel reinforced concrete composite structures using ensemble machine learning approaches

The bond-slip behavior between profiled steel and concrete is a critical issue in the steel reinforced concrete (SRC) composite structures. Existing models for predicting the bond strength of profiled steel–concrete were obtained based on small test data sets and may not provide accurate predictions when applied more broadly. To this end, this paper presents a data-driven analysis on the bond stress of profiled steel–concrete using ensemble machine learning approaches. A database of 249 individual tests in terms of bond stresses at different working conditions (i.e., bond stress at initial slip, ultimate bond stress, and residual bond stress) associating geometries, stirrup and profiled steel characteristics and concrete property were compiled from available literature. Based on the constructed database, six existing empirical models of ultimate bond stress were evaluated and the assessment results demonstrate that these models cannot achieve a desirable accuracy for their predictions. Then four independent ensemble machine learning algorithms were employed to develop the prediction models of bond stresses. The hyper-parameters of the developed models were optimized using Bayesian optimization method in order to improve their prediction performance. Finally, the prediction results of the ensemble machine learning-based models were interpreted using SHapley Additive exPlanations (SHAP) and feature importance analysis. The comprehensive comparisons demonstrate that jointly adopting ensemble machine learning and Bayesian optimization approaches can estimate the bond stress of profiled steel–concrete with high predictability and interpretability.

Street closure prediction based on the combined conditions of spatially collocated municipal infrastructure assets at the segment level

Generally, intervention activities that require excavations tend to create negative socio-economic impacts, such as increased traffic congestion and travel time, noise and air pollution, etc. A holistic form of intervention planning is needed to minimize the negative impacts and the frequency of maintenance, repair, and rehabilitation activities resulting in partial or complete street closures. This planning should be based on a unified classification model, which evaluates the conditions of different municipal assets (i.e., water and sewer pipes and road pavements) at the segment-level. The objectives of this paper are: (1) Developing a machine learning method for systematic condition classification of different spatially collocated underground municipal assets within a segment; and (2) Applying a heuristic approach for determining street closures based on the synchronized or unsynchronized interventions at the segment level induced from combining the interventions of individual assets within each segment. The conditions of all three assets are determined using three independent ensemble machine learning models, each made up of Random Forest, Gradient Boosted Trees, and Deep Learning as base learners. The required asset-level interventions are determined by these conditions combined with heuristics specific to each spatially collocated asset in a street segment to decide the segment-level synchronized or unsynchronized interventions and the nature of the street closures (partial, complete, or both) needed to undertake the interventions. The accuracies of the three models are 98.64%, 96.37%, and 82.38% for the pavement, water and sewer pipes, respectively. The segment-level intervention strategies resulted in both synchronized and unsynchronized interventions leading to complete, partial, or a combination of both partial and complete street closures. The predicted segment-level interventions had an accuracy of 79.92%. This research will aid municipal decision-makers in identifying and prioritizing synchronized segment-level interventions, therefore, improving the planning and estimation of the intervention durations and street closures.

Application-specific optimal model weighting of global climate models: A red tide example

Global climate models (GCMs) and Earth system models (ESMs) provide many climate services with environmental relevance. The High Resolution Model Intercomparison Project (HighResMIP) of the Coupled Model Intercomparison Project Phase 6 (CMIP6) provides model runs of GCMs and ESMs to address regional phenomena. Developing a parsimonious ensemble of CMIP6 requires multiple ensemble methods such as independent-model subset selection, prescreening-based subset selection, and model weighting. The work presented here focuses on application-specific optimal model weighting, with prescreening-based subset selection. As such, independent ensemble members are categorized, selected, and weighted based on their ability to reproduce physically-interpretable features of interest that are problem-specific. We discuss the strengths and caveats of optimal model weighting using a case study of red tide prediction in the Gulf of Mexico along the West Florida Shelf. Red tide is a common name of specific harmful algal blooms that occur worldwide, causing adverse socioeconomic and environmental impacts. Our results indicate the importance of prescreening-based subset selection as optimal model weighting can underplay robust ensemble members by optimizing error cancellation. Prescreening-based subset selection also provides insights about the validity of the model weights. By illustrating the caveats of using non-representative models when optimal model weighting is used, the findings and discussion of this study are pertinent to many other climate services.

Implicit ensemble tangent linear models (IETLMs) for model differentiation

AbstractIdeally, tangent linear models (TLMs) predict the difference between perturbed and unperturbed non‐linear forecasts of interest. The adjoint of a TLM gives the gradient of the non‐linear model and is used in 4DVar data assimilation and in adjoint‐based Forecast Sensitivity to Observation Impact (FSOI). The accuracy of the local ensemble TLM (LETLM) has been shown to be limited by its inability to account for implicit time stepping. Here we derive implicit ensemble TLMs (IETLMs) that, at most, require the number of independent ensemble members to be equal to the number of variables in the implicit computational stencil. The accuracy of the IETLM in the linear regime is confirmed using an implicitly time‐stepped Lorenz 96 model and a nine‐member ensemble. IETLMs feature two sparse matrices: matrix N that operates on an initially unknown future time perturbation, and matrix L that operates on the current time perturbation. For ensemble perturbations in the non‐linear regime, we develop a diagonally robust (DR) IETLM that reduces the chances of N becoming ill‐conditioned. The performance of the DR IETLM was compared with traditional TLM performance using IETLM ensemble perturbations whose ‘Gilmour et al., 2001’ measure of non‐linearity ranged up to the non‐linearity of operational 32‐hr ensemble forecast perturbations. Over a wide range of non‐linearity, the DR IETLM performance was found to match that of the traditional TLM provided the initial standard deviation of the ensemble perturbations was times the standard deviation of the test perturbations. Ideal FSOI requires the adjoint of a TLM that accurately predicts the known difference between corrected and uncorrected non‐linear forecasts. The DR IETLM was found to meet this FSOI accuracy requirement much more closely than the traditional TLM when the ensemble perturbations were created by subtracting the corrected forecast from ensemble members that were centred on the uncorrected forecast.

Prescreening-Based Subset Selection for Improving Predictions of Earth System Models With Application to Regional Prediction of Red Tide.

We present the ensemble method of prescreening-based subset selection to improve ensemble predictions of Earth system models (ESMs). In the prescreening step, the independent ensemble members are categorized based on their ability to reproduce physically-interpretable features of interest that are regional and problem-specific. The ensemble size is then updated by selecting the subsets that improve the performance of the ensemble prediction using decision relevant metrics. We apply the method to improve the prediction of red tide along the West Florida Shelf in the Gulf of Mexico, which affects coastal water quality and has substantial environmental and socioeconomic impacts on the State of Florida. Red tide is a common name for harmful algal blooms that occur worldwide, which result from large concentrations of aquatic microorganisms, such as dinoflagellate Karenia brevis, a toxic single celled protist. We present ensemble method for improving red tide prediction using the high resolution ESMs of the Coupled Model Intercomparison Project Phase 6 (CMIP6) and reanalysis data. The study results highlight the importance of prescreening-based subset selection with decision relevant metrics in identifying non-representative models, understanding their impact on ensemble prediction, and improving the ensemble prediction. These findings are pertinent to other regional environmental management applications and climate services. Additionally, our analysis follows the FAIR Guiding Principles for scientific data management and stewardship such that data and analysis tools are findable, accessible, interoperable, and reusable. As such, the interactive Colab notebooks developed for data analysis are annotated in the paper. This allows for efficient and transparent testing of the results' sensitivity to different modeling assumptions. Moreover, this research serves as a starting point to build upon for red tide management, using the publicly available CMIP, Coordinated Regional Downscaling Experiment (CORDEX), and reanalysis data.

The Brown measure of the sum of a self-adjoint element and an elliptic element

We completely determine the Brown measure of the sum of a self-adjoint element and an elliptic element, which is the limiting eigenvalue distribution of the random matrix YN+ s−t 2XN+i t 2XN′ where YN is an N×N deterministic Hermitian matrix whose eigenvalue distribution converges as N→∞ and XN and XN′ are independent Gaussian unitary ensembles. We also study various asymptotic behaviors of this Brown measure as the variance of the elliptic element approaches infinity.

Wavefunction structure in quantum many-fermion systems with k-body interactions: conditional q-normal form of strength functions

For finite quantum many-particle systems modeled with say m fermions in N single particle states and interacting with k-body interactions (k ⩽ m), the wavefunction structure is studied using random matrix theory. The Hamiltonian for the system is chosen to be H = H 0(t) + λV(k) with the unperturbed H 0(t) Hamiltonian being a t-body operator and V(k) a k-body operator with interaction strength λ. Representing H 0(t) and V(k) by independent Gaussian orthogonal ensembles of random matrices in t and k fermion spaces, respectively, the first four moments, in m-fermion spaces, of the strength functions F κ (E) are derived; the strength functions contain all of the information about wavefunction structure. With E denoting the H energies or eigenvalues and κ denoting unperturbed basis states with energy E κ , the F κ (E) give the spreading of the κ states over the eigenstates E. It is shown that the first four moments of F κ (E) are essentially the same as that of the conditional q-normal distribution given in (Szabowski 2010 Electron. J. Probab. 15 1296). This naturally gives asymmetry in F κ (E) with respect to E as E κ increases and also the peak value changes with E κ . Thus, the wavefunction structure in quantum many-fermion systems with k-body interactions in general follows the conditional q-normal distribution.

A Novel Method Based on Combination of Independent Component Analysis and Ensemble Empirical Mode Decomposition for Removing Electrooculogram Artifacts From Multichannel Electroencephalogram Signals

Electrooculogram (EOG) is one of common artifacts in recorded electroencephalogram (EEG) signals. Many existing methods including independent component analysis (ICA) and wavelet transform were applied to eliminate EOG artifacts but ignored the possible impact of the nature of EEG signal. Therefore, the removal of EOG artifacts still faces a major challenge in EEG research. In this paper, the ensemble empirical mode decomposition (EEMD) and ICA algorithms were combined to propose a novel EEMD-based ICA method (EICA) for removing EOG artifacts from multichannel EEG signals. First, the ICA method was used to decompose original EEG signals into multiple independent components (ICs), and the EOG-related ICs were automatically identified through the kurtosis method. Then, by performing the EEMD algorithm on EOG-related ICs, the intrinsic mode functions (IMFs) linked to EOG were discriminated and eliminated. Finally, artifact-free IMFs were projected to obtain the ICs without EOG artifacts, and the clean EEG signals were ultimately reconstructed by the inversion of ICA. Both EOGs correction from simulated EEG signals and real EEG data were studied, which verified that the proposed method could achieve an improved performance in EOG artifacts rejection. By comparing with other existing approaches, the EICA obtained the optimal performance with the highest increase in signal-to-noise ratio and decrease in root mean square error and correlation coefficient after EOG artifacts removal, which demonstrated that the proposed method could more effectively eliminate blink artifacts from multichannel EEG signals with less error influence. This study provided a novel promising method to eliminate EOG artifacts with high performance, which is of great importance for EEG signals processing and analysis.

Condition Monitoring of Slow-speed Gear Wear using a Transmission Error-based Approach with Automated Feature Selection

Gear flank changes caused by wear do not only affect the dynamic behavior of gear systems, but they can also compromise the load-carrying capacity of gear teeth up to critical failure. To help avoid unintended consequences like downtime or safety risks, a condition monitoring system needs to be able to estimate the current wear during operation based on available sensor measurements. While many condition monitoring approaches in research rely on vibrational analysis with manual feature engineering, gearboxes running at slow speed do not reveal much excitation information for this purpose. We therefore introduce an approach for slow-speed gear wear monitoring that is based on the dynamic gear transmission error and that contains an automated feature selection process. For this purpose, we extract a large set of features from the preprocessed transmission error samples. Applying combined filter and embedded feature selection methods enables us to automatically identify and remove features with low relevance. The selection process consists of filtering features with no statistical dependence on the target wear value, removing redundant features with a correlation analysis and a recursive feature elimination process with cross-validation based on a random forest regressor. The remaining relevant set of features is the basis for model training and subsequent wear estimation. For this, the present research employed two independent ensemble models, random forest regression and gradient boosted regression trees. To train and test the proposed approach, we conducted slow-speed gear experiments with developing gear wear on a single-stage spur gear test rig setup. The results of both models show good gear wear estimation performance compared to the actual wear mass loss, even for small quantities. Hence, the proposed transmission error-based approach with automated feature selection is able to quantify the degree of slow-speed wear and offers a possible way for condition monitoring and fault diagnosis.

A 10 per cent increase in global land evapotranspiration from 2003 to 2019.

Accurate quantification of global land evapotranspiration is necessary for understanding variability in the global water cycle, which is expected to intensify under climate change1-3. Current global evapotranspiration products are derived from a variety of sources, including models4,5, remote sensing6,7 and in situ observations8-10. However, existing approaches contain extensive uncertainties; for example, relating to model structure or the upscaling of observations to a global level11. As a result, variability and trends in global evapotranspiration remain unclear12. Here we show that global land evapotranspiration increased by 10±2 per cent between 2003 and 2019, and that landprecipitation is increasingly partitioned into evapotranspiration rather than runoff. Our results are based on an independent water-balance ensemble time series of global land evapotranspiration and the corresponding uncertainty distribution, using data from the Gravity Recovery and Climate Experiment (GRACE) and GRACE-Follow On (GRACE-FO) satellites13. Variability in global land evapotranspiration is positively correlated with El Niño-Southern Oscillation. The main driver of the trend, however, is increasing land temperature. Our findings provide an observational constraint on globalland evapotranspiration, and are consistent with the hypothesis that global evapotranspiration should increase in a warming climate.

INDEPENDENT ENSEMBLES AND NEW DIRECTIONS IN THEATRICAL ART OF BELARUS OFTHE 2010S

The paper briefly describes creative activity of Belarusian independent theatre ensembles in experimental genres of stage art in the 2010s. Range of creative quest of Belarusian theatre of the second decade of the 21st century embraces various genres and directions: social, documentary, inclusive theatre, interactive baby theatre, puppet show, etc. The author points out on creative approach of Belarusian independent theatre figures to understanding and adaptation of conceptual and artistic achievement of directors' and actors' experiences f the world theatre, that exclude its blind taking.

On mesofractal interpretation of the galaxy spatial arrangement

The term mesofractal is used here for a system of random points arrangement of which reveals the fractal character on small scales and looks as uniform on large scales. The way of simulation of such systems is based on the random jump process with Lévy-Feldheim stable distributions taken as transition densities, and the Poisson point process for the origin points of the independent trajectory ensemble. Such model successfully represents main statistical properties of the observed part of the universe and may be useful for testing estimation procedures. This article contains also a short discussion of evolution of cell-count data between fractal and uniform domains.

A heat transfer model of fully developed turbulent channel flow

Experimental and numerical studies over the past two decades indicate that as the Reynolds number becomes large the turbulent boundary layer is increasingly composed of zones of uniform streamwise momentum, segregated by narrow regions of high shear. Recent experimental evidence suggests that passive scalar fields (for example, temperature) in turbulent boundary layers at high Reynolds number show similar characteristics; namely, large uniform temperature zones (UTZs) separated by narrow regions of high gradient, which we term thermal fissures (TFs). Herein, a model informed by analysis of the mean scalar transport equation, and that leverages the dynamical model recently developed by the authors (Cuevas Bautista et al., J. Fluid Mech., vol. 858, 2019, pp. 609-633), is formulated to predict passive scalar transport using the UTZ/TF concept. First, a finite number of TFs are distributed across the boundary layer. In analogy with the aforementioned dynamical model, the wall-normal positions of the TFs and their characteristic temperatures are then perturbed to generate independent ensembles, from which statistical moments are computed. The model successfully reproduces the statistical profiles of the temperature field as well as the streamwise turbulent heat flux. Lastly, the Prandtl number dependency of the empirically chosen parameters is investigated. It is concluded that the higher-order statistics, especially the kurtosis, produced by the model are sensitive to the Prandtl number, while the mean temperature and turbulent heat flux do not show noticeable Prandtl number dependency.

Synchronization of Independent Neural Ensembles in Human EEG during Choice Tasks.

During behavioral experiments, humans placed in a situation of having to choose between a more valuable but risky reward and a less valuable but guaranteed reward make their decisions in accordance with external situational factors and individual characteristics, such as inclination to risk or caution. In such situations, humans can be divided into “risk-inclined” and “risk-averse” (or “cautious”) subjects. In this work, characteristics of EEG rhythms, such as phase–phase relationships and time lags between rhythms, were studied in pairs of alpha–beta and theta–beta rhythms. Phase difference can also be expressed as a time lag. It has been suggested that statistically significant time lags between rhythms are due to the combined neural activity of anatomically separate, independent (in activation/inhibition processes) ensembles. The extents of synchronicity between rhythms were compared as percentages between risk-inclined and risk-averse subjects. The results showed that synchronicity in response to stimuli was more often observed in pairs of alpha–beta rhythms of risk-averse subjects compared with risk-inclined subjects during the choice of a more valuable but less probable reward. In addition, significant differences in the percentage ratio of alpha and beta rhythms were revealed between (i) cases of synchronization without long time lags and (ii) cases with long time lags between rhythms (from 0.08 to 0.1 s).

Improving ADA-boost as a Popular Ensemble in Classification Problems

In Data Mining, several classification algorithms are used to perform classification based on single learner but classification accuracy is not in an effective manner. To increase the accuracy of the classification then multiple learners are combined to get better results. The multiple learners are trained and combined into an ensemble. The ensemble can increase generalization ability and robustness [3]. Based on the advantage of ensemble, the ensemble classification is a major concern in research directions of machine learning. Another importance of ensemble is that it is much stronger than single base learner to produce accurate hypothesis. The ensembles are divided into homogeneous or heterogeneous, dependent or independent ensembles. The dependent ensemble methods like boosting and AdaBoost algorithms are promisingly provide an accurate hypothesis. Finally, AdaBoost can be a better classifier ensemble to generate accurate results.