Model-based Sampling Research Articles

Using group level factor models to resolve high dimensionality in model-based sampling.

Joint modeling of decisions and neural activation poses the potential to provide significant advances in linking brain and behavior. However, methods of joint modeling have been limited by difficulties in estimation, often due to high dimensionality and simultaneous estimation challenges. In the current article, we propose a method of model estimation that draws on state-of-the-art Bayesian hierarchical modeling techniques and uses factor analysis as a means of dimensionality reduction and inference at the group level. This hierarchical factor approach can adopt any model for the individual and distill the relationships of its parameters across individuals through a factor structure. We demonstrate the significant dimensionality reduction gained by factor analysis and good parameter recovery, and illustrate a variety of factor loading constraints that can be used for different purposes and research questions, as well as three applications of the method to previously analyzed data. We conclude that this method provides a flexible and usable approach with interpretable outcomes that are primarily data-driven, in contrast to the largely hypothesis-driven methods often used in joint modeling. Although we focus on joint modeling methods, this model-based estimation approach could be used for any high dimensional modeling problem. We provide open-source code and accompanying tutorial documentation to make the method accessible to any researchers. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

PRISM: A Hierarchical Intrusion Detection Architecture for Large-Scale Cyber Networks

The increase in scale of cyber networks and the rise in sophistication of cyber-attacks have introduced several challenges in intrusion detection. The primary challenge is the requirement to detect complex multi-stage attacks in realtime by processing the immense amount of traffic produced by present-day networks. In this paper we present PRISM, a hierarchical intrusion detection architecture that uses a novel attacker behavior model-based sampling technique to minimize the realtime traffic processing overhead. PRISM has a unique multi-layered architecture that monitors network traffic distributedly to provide efficiency in processing and modularity in design. PRISM employs a Hidden Markov Model-based prediction mechanism to identify multi-stage attacks and ascertain the attack progression for a proactive response. Furthermore, PRISM introduces a stream management procedure that rectifies the issue of alert reordering when collected from distributed alert reporting systems. To evaluate the performance of PRISM, multiple metrics have been proposed, and various experiments have been conducted on a multi-stage attack dataset. The results exhibit up to 7.5x improvement in processing overhead as compared to a standard centralized IDS without the loss of prediction accuracy while demonstrating the ability to predict different attack stages promptly.

Time-dependent non-linear buckling of 3D CFST arch structures with hybrid random interval uncertainties

Limit point and bifurcation buckling loads are critical concerns in structural stability design. With the inevitable viscoelastic effects of creep and shrinkage in concrete, such critical buckling points may vary due to the time-dependent change of equilibrium configuration. Furthermore, the intrinsic uncertainty and natural randomness in the geometry and material characteristics would affect the structural stability performance significantly. The present study provides a new robust method, called the generalized Chebyshev surrogate model-based sampling approach, in assessing the time-dependent nonlinear buckling behaviour of 3D concrete-filled steel tubular (CFST) arch structures when both random and interval uncertainties are involved. In the proposed approach, the relationships between the uncertain parameters and the critical nonlinear limit and bifurcation buckling loads are formulated using Chebyshev surrogate model strategy combined with finite element method. The extreme bounds of the statistical features, including means, standard deviations, of the critical nonlinear buckling loads are furnished by using Monte Carlo method and Quasi Monte Carlo simulation method. Finally, the applicability and the validity of the proposed approach are illustrated with a series of numerical investigations.

Evaluation of several recently developed sampling strategies within the coarse pixel scale for validation of coarse-resolution satellite albedo products

ABSTRACT Due to the spatial heterogeneity and the spatial scale mismatch between in situ and satellite-based measurements, optimal ground sampling should be made to increase the representativeness of in situ observations. Therefore, many ground sampling strategies have been proposed, but their performance within the coarse pixel has not been evaluated. Hence, this study evaluated four typical methods regarding their ability to obtain pixel scale ground ‘truth’. Random combination (RC) performs best, with the always fewest samples to satisfy representativeness errors (REs) of 3% in the case of a small number of samples. When the goal of sampling is to obtain in situ measurements with REs close to 0 at the expense of increasing the number of samples, cumulative representativeness sampling (CRS) is more effective than RC in less heterogeneous areas. Geo-statistical model-based sampling (GSS) does not work well because the number of samples within the coarse pixel scale cannot support a robust semi-variogram model. Stratified sampling (SS) is highly dependent on spatial heterogeneity and does not work well in the case of small sample sizes. This study gives important guidance for ground sample deployment within the coarse pixel for validation of coarse-resolution satellite albedo products over a heterogeneous surface.

Taxon and trait-based sampling curves can be used as a tool for assessing impairment in salinized headwater streams

Many ecosystems are losing biodiversity, raising concern for the services they provide. However, the extent of loss is uncertain, especially for diverse insects, because of incomplete sampling. Modeling techniques for estimating upper bounds on diversity are needed to assess benthic freshwater insect abundances, taxa richness, and diversity because some species are declining at alarming rates while others are increasing. In central Appalachian streams of the U.S.A., aquatic insect communities have lost diversity following salinization caused by mining activities. However, the number of taxa observed are dependent upon sampling effort. Incomplete sampling could misrepresent biodiversity and functional losses. Our goal was to use taxon sampling curves to estimate sampling effort required to maximize the probability of accurate benthic macroinvertebrate characterization in dominant riffle habitats of headwater streams. We collected 5 to 10 quantitative benthic macroinvertebrate samples in each of six, first-order streams in the central Appalachian region. For our single-habitat, mesoscale approach, we predicted: (1) macroinvertebrate taxa richness would be the most robust indicator of salinization response compared to diversity, evenness, and density, (2) less sampling effort would be needed to capture taxonomic richness in salinized streams compared to reference streams, and (3) response diversity would also be lower in salinized streams because select trait states would be represented by fewer taxa. Results suggested nominally lower taxa richness, evenness, and diversity in some salinized streams but not in all despite greater or more variable within-stream densities. Nonetheless, sampling effort required to characterize macroinvertebrate communities did not differ between reference and salinized streams, though uneven within-stream taxa distributions corresponded with greater sampling effort requirements for complete characterization. Benthic macroinvertebrate community characteristics were different where higher densities and richness of small-bodied and fast-developing taxa were more common in salinized streams. Response diversity depended on trait states. For example, only about five shredder taxa were represented in salinized streams compared to eight taxa in reference streams. Despite some indication of lower response diversity across some functional feeding groups, more than five samples were needed for robust comparisons. Taxon and trait-based sampling curves suggest that greater overall sampling effort is needed and equal samples per stream was critical for complete diversity assessments regardless of the level of mining-induced salinization. Model-based sampling curves can serve as a tool to assess upper bounds on diversity metrics and sample-effort rigor.

Enhanced Adaptive Kriging Combined Adaptive Radial-Based Importance Sampling Method for Reliability Analysis

The surrogate model-based sampling method has been an active area of research due to its evident efficiency over direct sampling method. The fusion of the adaptive kriging (AK) model with the adaptive radial-based importance sampling (ARBIS) method is developed as an improvement of the AK model combined with Monte Carlo simulation (MCS) method especially for estimating the small failure probabilities. In the AK–ARBIS method, the number of candidate samples in each updating process of the kriging model is reduced through partitioning the whole MCS sample domain into different subdomains and removing the samples inside the last subdomain from the participating candidate sampling pool, which can avoid the time-consuming matrix operation involved in kriging predictions in order to find the best training sample in each updating step. In the existing AK–ARBIS method, kriging model stops updating according to the learning function-based stopping criterion, which cannot directly reflect the error of the estimated failure probability. In this regard, this paper proposes the error-based stopping criterion (ESC) enhanced AK–ARBIS method, where the processes of finding the optimal hypersphere and accurately identifying the sign of candidate samples are decoupled in order to successfully embed the ESC. In addition to embedding the ESC into the AK–ARBIS method, two target-oriented candidate sampling reduction strategies are investigated, respectively, to improve the efficiency of finding the optimal hypersphere and accurately identifying the sign of the participating candidate samples. Based on the candidate sampling pool reduction strategies and ESC, both the computational time and the required number of calls to the real limit state function are reduced over the existing AK–ARBIS method. Three case studies are analyzed to demonstrate the accuracy and efficiency of the proposed enhanced AK–ARBIS method, especially for estimating the small failure probabilities.

A novel multi-stage ensemble model for credit scoring based on synthetic sampling and feature transformation

Credit scoring has become increasingly important for financial institutions. With the advancement of artificial intelligence, machine learning methods, especially ensemble learning methods, have become increasingly popular for credit scoring. However, the problems of imbalanced data distribution and underutilized feature information have not been well addressed sufficiently. To make the credit scoring model more adaptable to imbalanced datasets, the original model-based synthetic sampling method is extended herein to balance the datasets by generating appropriate minority samples to alleviate class overlap. To enable the credit scoring model to extract inherent correlations from features, a new bagging-based feature transformation method is proposed, which transforms features using a tree-based algorithm and selects features using the chi-square statistic. Furthermore, a two-layer ensemble method that combines the advantages of dynamic ensemble selection and stacking is proposed to improve the classification performance of the proposed multi-stage ensemble model. Finally, four standardized datasets are used to evaluate the performance of the proposed ensemble model using six evaluation metrics. The experimental results confirm that the proposed ensemble model is effective in improving classification performance and is superior to other benchmark models.

Digital Implementation of Single Current Sensor-Based IPMSM Drive With Sampling Error Delay Compensation

Single current sensor-based three-phase drives are gaining huge interest from industries for sensitivity mismatch issues and cost reduction. However, the current sampling error caused by multiple sampling points during current reconstruction from various single current sensor (SCS) topologies led to increase in speed fluctuation and torque ripple. In this paper, a computationally efficient and accurate model-based sampling error compensation scheme based on a new SCS with double branch current sampling technique is proposed for an interior permanent magnet synchronous motor (IPMSM) drive. In order to improve of current reconstruction precision, the compensation strategy includes rotor angle compensation of IPMSM and dead-time effect mitigation for reduction of non-ideal behavior of inverter, which does not require major computation. Also, dc-link voltage utilization is improved significantly without any external dc-link control. The efficacy of the proposed compensation strategy is verified by an experimental setup consisting of 1.5 kW IPMSM drive, which shows the effectiveness of proposed current reconstruction technique from SCS by reducing current sampling error with minimum computational burden.

A semi-parametric empirical likelihood approach for conditional estimating equations under endogenous selection

The estimation and inference for conditional estimating equations models with endogenous selection, are considered. The approach takes into account possible endogenous selection which may lead to a selection bias. It can be used for a wide range of statistical models not covered by the model-based sampling theory. Endogeneity can be either part of the selection or within the covariates. It is particularly well suited for models with unknown heteroscedasticity, uncontrolled confounders and measurement errors. It will not be necessary to model the relationship between the endogenous covariates and the instrumental variables, which offers major advantages over two-stage least-squares. The approach proposed has the advantage of being based on a fixed number of constraints determined by the size of the parameter.

Robust Predictive Control for Modular Solid-State Transformer With Reduced DC Link and Parameter Mismatch

This article presents the analysis and implementation of a predictive control method for dc-link regulation and voltage balance in a cascaded modular reduced dc-link solid-state transformer (SST). Passive components like bulky dc links limit the power density of power converters, especially medium-voltage (MV) SST. Reduced dc-link or low-inertia converters can dramatically reduce the size, cost, and weight by tolerating larger dc-link ripples and improve the reliability with electrolytic capacitor-less dc link. However, a small dc link leads to tight coupling between the input and the output stages, which is a challenge for control design. In stacked low-inertia converters (SLIC), the low-inertia converter modules are stacked for MV applications, resulting in coupling between the modules and making the control more challenging. A new model predictive control method that can achieve deadbeat regulation on the dc link without weighting factors has been proposed to address this novel problem. This article focuses on analyzing the condition of the low-inertia dc link up to 80% ripple, the robustness of the control under parameter mismatches, high-order terms, and important implementation issues, such as model-based sampling and computation delay compensation. Significantly, the high-order terms are introduced because of the large dc-link ripple. These high-order terms are unique to the SLIC and negligible in conventional high-inertia converters. A discrete-time large-signal model is built to capture the dc-link's nonlinear dynamics, and the eigenvalues of a small-signal Jacobian matrix are analyzed with Floquet theory to evaluate stability, using the modular soft-switching SST (M-S4T) as an example of the SLIC. Simulation and experimental results of an MVDC M-S4T verify the analysis and the predictive control method. Finally, the general application of the predictive control to low-inertia converters is compared against a conventional PI controller using a reduced dc-link active-front-end rectifier as an example.

Kenneth (Ken) Ronald Walter Brewer (1931–2021)

Kenneth (Ken) Ronald Walter Brewer (1931–2021)

An efficient dimensionality-independent algorithm for failure probability-based global sensitivity analysis by dual-stage adaptive kriging model

The failure probability-based global sensitivity index (FPGSI) analyses how the model inputs affect the failure probability of a model. It is useful for guiding reliability-based design optimization and enhancing reliability by controlling the uncertainty of the important input variables. Based on the law of total variance in successive intervals without overlapping and the dual-stage adaptive kriging (AK) model-based importance sampling (IS) method, an efficient dimensionality-independent method is proposed. First, an interval-conditional failure probability-based formula is established. Secondly, a dual-stage AK model is embedded into the formula to construct the IS probability density function and identify the state (failed or safe) of every IS sample. Thirdly, using different partitions of IS samples, all inputs’ FPGSIs can be simultaneously obtained by taking the corresponding subdomains’ samples into the proposed computational formula. The results of four case studies illustrate the effectiveness of the proposed algorithm, especially for cases with multiple failure regions.

Model-based Respondent-driven sampling analysis for HIV prevalence in brazilian MSM

Respondent Driven Sampling study (RDS) is a population sampling method developed to study hard-to-reach populations. A sample is obtained by chain-referral recruitment in a network of contacts within the population of interest. Such self-selected samples are not representative of the target population and require weighing observations to reduce estimation bias. Recently, the Network Model-Assisted (NMA) method was described to compute the required weights. The NMA method relies on modeling the underlying contact network in the population where the RDS was conducted, in agreement with directly observable characteristics of the sample such as the number of contacts, but also with more difficult-to-measure characteristics such as homophily or differential characteristics according to the response variable. Here we investigated the use of the NMA method to estimate HIV prevalence from RDS data when information on homophily is limited. We show that an iterative procedure based on the NMA approach allows unbiased estimations even in the case of strong population homophily and differential activity and limits bias in case of preferential recruitment. We applied the methods to determine HIV prevalence in men having sex with men in Brazilian cities and confirmed a high prevalence of HIV in these populations from 3.8% to 22.1%.

Modeling algorithm influence on the success of predicting new populations of rare species: ground-truthing models for the Pale-Belly Frost Lichen (Physconia subpallida) in Ontario

Conservation of rare species relies on a thorough knowledge of distributions and the locations of populations. Field surveys for new populations of rare species are time consuming and resource intensive. Thus, it is essential to conduct searches as efficiently as possible. Model-based sampling, where species distribution models are used to select sites with high probability of the species presence for surveys, can drastically improve the efficiency and success of field sampling. The vast array of methods to build species distribution models can make it difficult to select one approach to implement for a project. Here we directly compared two methods, Maxent and non-parametric multiplicative regression (NPMR), using the endangered lichen Physconia subpallida as the focal species. We built models using all known localities of P. subpallida in Ontario, Canada, then ground-truthed each of the models for 9 days over a 2-year period, searching only areas predicted to have the highest level of probability of species occurrence. NPMR far outperformed Maxent, with the discovery of six new populations with a total of 36 individuals compared to one new population consisting of a single individual, respectively. The disparity between the two results likely stems from the potentially over-simplified response curves from Maxent when compared to NPMR. If a complex relationship is expected between the species and environmental variables, NPMR may outperform Maxent. However, there is not a single modeling algorithm that works best in every situation, so it is essential to test multiple modeling methods for guiding rare species surveys.

Improved spatial ecological sampling using open data and standardization: an example from malaria mosquito surveillance.

Vector-borne disease control relies on efficient vector surveillance, mostly carried out using traps whose number and locations are often determined by expert opinion rather than a rigorous quantitative sampling design. In this work we propose a framework for ecological sampling design which in its preliminary stages can take into account environmental conditions obtained from open data (i.e. remote sensing and meteorological stations) not necessarily designed for ecological analysis. These environmental data are used to delimit the area into ecologically homogeneous strata. By employing Bayesian statistics within a model-based sampling design, the traps are deployed among the strata using a mixture of random and grid locations which allows balancing predictions and model-fitting accuracies. Sample sizes and the effect of ecological strata on sample sizes are estimated from previous mosquito sampling campaigns open data. Notably, we found that a configuration of 30 locations with four households each (120 samples) will have a similar accuracy in the predictions of mosquito abundance as 200 random samples. In addition, we show that random sampling independently from ecological strata, produces biased estimates of the mosquito abundance. Finally, we propose standardizing reporting of sampling designs to allow transparency and repetition/re-use in subsequent sampling campaigns.

Expected Changes of Montenegrin Climate, Impact on the Establishment and Spread of the Asian Tiger Mosquito (Aedes albopictus), and Validation of the Model and Model-Based Field Sampling

Aedes albopictus has become established in many parts of Europe since its introduction at the end of the 20th century. It can vector a range of arboviruses, of which Chikungunya and Dengue are most significant for Europe. An analysis of the expected climate change and the related shift in Köppen zones for Montenegro and impact on the establishment of Ae. albopictus was conducted. Outputs of a mechanistic Aedes albopictus model were validated by 2245 presence/absence records collected from 237 different sites between 2001 and 2014. Finally, model-based sampling was designed and performed at 48 sites in 2015, in a previously unexplored northern part of Montenegro, and results were validated. The Eta Belgrade University (EBU)-Princeton Ocean Model (POM) regional climate model was used with the A2 emissions scenario for the 2001–2030 and 2071–2100 integration periods. The results point to a significant increase in suitability for the mosquito and a vertical shift to higher altitudes by the end of the century. The model showed excellent results with the area under the receiver operating characteristic curve (AUC) of 0.94. This study provides a tool for prioritizing surveillance efforts (model-based surveillance), especially when resources are limited. This is the first published analysis of Climate Change that incorporates observations from the national synoptic grid and the subsequent impact on Ae. albopictus in Montenegro.

Tumor Cell Load and Heterogeneity Estimation From Diffusion-Weighted MRI Calibrated With Histological Data: an Example From Lung Cancer.

Diffusion-weighted magnetic resonance imaging (DWI) is a key non-invasive imaging technique for cancer diagnosis and tumor treatment assessment, reflecting Brownian movement of water molecules in tissues. Since densely packed cells restrict molecule mobility, tumor tissues produce usually higher signal (a.k.a. less attenuated signal) on isotropic maps compared with normal tissues. However, no general quantitative relation between DWI data and the cell density has been established. In order to link low-resolution clinical cross-sectional data with high-resolution histological information, we developed an image processing and analysis chain, which was used to study the correlation between the diffusion coefficient (D value) estimated from DWI and tumor cellularity from serial histological slides of a resected non-small cell lung cancer tumor. Color deconvolution followed by cell nuclei segmentation was performed on digitized histological images to determine local and cell-type specific 2d (two-dimensional) densities. From these, the 3d cell density was inferred by a model-based sampling technique, which is necessary for the calculation of local and global 3d tumor cell count. Next, DWI sequence information was overlaid with high-resolution CT data and the resected histology using prominent anatomical hallmarks for co-registration of histology tissue blocks and non-invasive imaging modalities' data. The integration of cell numbers information and DWI data derived from different tumor areas revealed a clear negative correlation between cell density and D value. Importantly, spatial tumor cell density can be calculated based on DWI data. In summary, our results demonstrate that tumor cell count and heterogeneity can be predicted from DWI data, which may open new opportunities for personalized diagnosis and therapy optimization.

Test of sampling methods to optimize the calibration of vine water status spatial models

Plant water status is one of the main factors affecting yield and quality in viticulture. Nevertheless, it is generally difficult to characterize it with enough precision for management purposes. In addition to its temporal variation, related to climate conditions, it has been shown that it is also spatially variable within the vineyard. In practical terms, this makes traditional reference measurements both too costly and time consuming to be affordable. In contrast, it has been shown that spatial variation of plant water status can be inferred from more accessible information, such as plant vigour in Mediterranean conditions. The main practical limitation for this approach is that the relationship between vigour measurements and plant water status is specific for each block and needs to be explicitly calibrated. Furthermore, a high number of measurements are usually required for this calibration. The objective of this work was to propose and test sampling methods to optimize the calibration of a specific spatial model of vine water status using the minimum number of measurements. Two model-based sampling methods commonly used in non-spatial modelling, Kennard and Stone (K&S) and Surface Response (SR) were considered, tested and discussed. Satisfactory results were obtained with both methods: with a sample size of 9 calibration sites, both sampling methods gave similar errors to the reference model (Root Mean Standard Error of Prediction, RMSEP = 0.1 MPa), which was calibrated with 49 sites. Taking into consideration the advantages and limitations of each method, K&S is considered to be better adapted for the case study presented. The proposed sampling approach could be extended to other spatial models used in precision agriculture in which ancillary variables can be used to explain most of the spatial variation for any agronomic information of interest.

Mapping Soil Organic Carbon stocks and estimating uncertainties at the regional scale following a legacy sampling strategy (Southern Belgium, Wallonia)

The quantification and the spatialisation of reliable SOC stocks (MgCha−1) and total stock (TgC) baselines and associated uncertainties are fundamental to detect the gains or losses in SOC, and to locate sensitive areas with low SOC levels. Here, we aim to both quantify and spatialize SOC stocks at regional scale (southern Belgium) based on data from one non-design-based or model-based sampling scheme. To this end, we developed a computation procedure based on Digital Soil Mapping techniques and stochastic simulations (Monte-Carlo) allowing the estimation of multiple (here, 10,000) independent spatialized datasets. The computation of the prediction uncertainty accounts for the errors associated to both the estimations of i) SOC stocks and ii) parameters of the spatial model. Based on these 10,000 individuals, median SOC stocks and 90% prediction intervals were computed for each pixel, as well as total SOC stocks and their 90% prediction intervals for selected sub-areas and for the entire study area. Hence, a Generalised Additive Model (GAM) explaining 69.3% of the SOC stock variance was calibrated and then validated (R2=0.64). The model overestimated low SOC stock (below 50MgCha−1) and underestimated high SOC stock (especially those above 100MgCkg−1). A positive gradient of SOC stock occurred from the northwest to the center of Wallonia with a slight decrease on the southernmost part, correlating to the evolution of precipitation and temperature (along with elevation) and dominant land use. At the catchment scale higher SOC stocks were predicted on valley bottoms, especially for poorly drained soils under grassland. Mean predicted SOC stocks for cropland and grassland in Wallonia were of 26.58TgC (SD 1.52) and 43.30TgC (2.93), respectively. The procedure developed here allowed to predict realistic spatial patterns of SOC stocks all over agricultural lands of southern Belgium and to produce reliable statistics of total SOC stocks for each of the 20 combinations of land use/agricultural regions of Wallonia. This procedure appears useful to produce soil maps as policy tools in conducting sustainable management at regional and national scales, and to compute statistics which comply with specific requirements of reporting activities.

Predicting Logging Residue Volumes in the Pacific Northwest

Pacific Northwest forest managers seek estimates of post-timber-harvest woody residue volumes and biomass that can be related to readily available siteand tree-level attributes. To better predict residue production, researchers investigated variability in residue ratios, growing-stock residue volume per mill-delivered volume, across Idaho, Montana, Oregon, and Washington. This project presented unique sample design challenges, and the authors adopted model-based sampling to calculate the growing-stock logging residue ratio for the four-state region and produced models that relate the residue ratio to individual treeand stand-level variables meaningful to land managers. The regionwide residue ratio was 0.0269, i.e., 26.9 ft of growing-stock logging residue per 1,000 ft (26.9 m per 1,000 m) of mill-delivered volume. Residue ratios were related to treeand site-level variables with predictive models. Residue ratios were predicted to increase with larger small-end used diameter and decline exponentially with increasing dbh. Ratios were predicted to drop when pulp logs were removed and when timber was mechanically felled. Results from this study could be used to produce or improve residue prediction tools for land managers.