Efficient Sampling Scheme Research Articles

P015 Detection of endoscopic lesions from limited quality annotations in colonoscopy videos

Abstract Background Ulcerative colitis is an idiopathic inflammatory disorder affecting the mucosa of the colon with superficial erosion and ulcers associated with bleeding. Severity assessment using current scoring schemes such as UCEIS and MAYO relies on the subjective interpretation of the physician and fails to take into account the size of the lesions, their number and distribution. Automatic lesion detection methods can enable fine-grained assessment of lesion severity, but require training stage based on time-consuming manual annotation. Most methods currently use generic datasets that are biased towards capsule endoscopy and are not adapted to locally available hardware. Methods We learn automatic bleeding and ulcer detectors on a local data set created by the Gastroenterology group at the Bichat and Beaujon hospitals. The patients’ videos were anonymous, analysed after obtaining their consent. The study was approved by the local research study committee. To minimise expert annotation burden, only rectangular annotations are provided instead of a precise delineation of lesion boundaries (Figure 1). This leads to many mislabelled pixels, especially in the corners, and affects the evaluation of the models’ performance and our ability to find correct models. Standard sensitivity and specificity cannot be used effectively on this data-set. We propose to evaluate model sensitivity on the annotation level and keep specificity at the pixel level. On the training set, we consider that a model correctly identifies a lesion if it agrees with the expert on a subset of the annotation, and count the detected annotations weighted by their area. For robustness and ease of interpretation, we explore the set of linear classifiers, and propose an efficient sampling scheme that rejects trivial models. This method is evaluated on a database of 10 colonoscopy videos (5 training videos and 5 test videos). Results In spite of the limited quality of the annotations, we find lesion detectors with a good annotation-level sensitivity (Table 1) and visual performance (see Figure 1). The detector’s performance is computed reliably. We evaluate sensitivity and specificity on 20 random subsets containing 10\% of the images, and obtain similar performance for the same patient for all models (cf Figure 2). Conclusion Despite mislabeled pixels, we obtain lesion detectors with good performance, and we show that the performance is computed reliably. However, the inter-patient performance is variable and the best models fail on some patients (sensitivity below 20\% in some cases Figure 2). This suggests that the models are not universal and that the appearance of bleeding and ulcers should be normalised further before automatic detection.

AUTOMATIC BLEEDING AND ULCER DETECTION FROM LIMITED QUALITY ANNOTATIONS IN ULCERATIVE COLITIS

Ulcerative colitis is an idiopathic inflammatory disorder affecting the mucosa of the colon with superficial erosions and ulcers associated with bleeding. Severity assessment using current scoring schemes such as UCEIS and MAYO relies on the subjective interpretation of the physician and fails to take into account the size of the different lesions, their number and their distribution throughout the colon. Automatic lesion detection and grading procedures can enable fine-grained assessment of lesion severity for treatment follow-up. This work aims to learn automatic bleeding and ulcer lesion detectors. As such algorithms may need to be tuned to the characteristics of the equipment used to perform the colonoscopy, we train our detectors on a dataset obtained by the Gastroenterology group at the Bichat and Beaujon hospitals. The patients' videos were anonymous, analyzed after obtaining their consent. The study was approved by the local research study committee. To minimize expert annotation burden, only rectangular annotations are required instead of a precise delineation of lesion boundaries (Figure 1). However, this dataset contains many mislabeled pixels, especially in the corners of the rectangles. This affects the evaluation of the models' performance and our ability to find correct models. Standard sensitivity and specificity cannot be used effectively on this dataset. We propose to evaluate model sensitivity on the annotation level and keep specificity at the pixel level. On the training set, we consider that a model correctly identifies a lesion if it agrees with the expert on a subset of the annotation, and count the detected annotations weighted by their area. For robustness and ease of interpretation, we explore the set of linear classifiers, and propose an efficient sampling scheme that rejects trivial models. This method is evaluated on a database of 10 colonoscopy videos (5 training videos and 5 test videos). In spite of the limited quality of the annotations, we find lesion detectors with a good annotation-level sensitivity (93% specificity / 89% sensitivity for bleeding and 57% specificity / 83% sensitivity for ulcers) and visual performance (see Figure 1). The detector performance is computed reliably. We evaluated sensitivity and specificity on 20 random subsets containing 10% of the images, and obtain similar performance for the same patient for all models (cf Figure 2). However, we observe that the inter-patient performance is variable and that the best models can fail on some patients. In Figure 2, the sensitivity is below 20% is some cases. This suggests that the models are not universal because bleeding and ulcers have a different appearance in patients, and that this should be corrected before automatic detection. Alternatively, models should be adapted to the patient characteristics.

Cross-classes domain inference with network sampling for natural resource inventory

There are two distinct types of domains, design- and cross-classes domains , with the former extensively studied under the topic of small-area estimation. In natural resource inventory, however, most classes listed in the condition tables of national inventory programs are characterized as cross-classes domains, such as vegetation type, productivity class, and age class. To date, challenges remain active for inventorying cross-classes domains because these domains are usually of unknown sampling frame and spatial distribution with the result that inference relies on population-level as opposed to domain-level sampling. Multiple challenges are noteworthy: (1) efficient sampling strategies are difficult to develop because of little priori information about the target domain; (2) domain inference relies on a sample designed for the population, so within-domain sample sizes could be too small to support a precise estimation; and (3) increasing sample size for the population does not ensure an increase to the domain, so actual sample size for a target domain remains highly uncertain, particularly for small domains. In this paper, we introduce a design-based generalized systematic adaptive cluster sampling (GSACS) for inventorying cross-classes domains. Design-unbiased Hansen-Hurwitz and Horvitz-Thompson estimators are derived for domain totals and compared within GSACS and with systematic sampling (SYS). Comprehensive Monte Carlo simulations show that (1) GSACS Hansen-Hurwitz and Horvitz-Thompson estimators are unbiased and equally efficient, whereas the latter outperforms the former for supporting a sample of size one; (2) SYS is a special case of GSACS while the latter outperforms the former in terms of increased efficiency and reduced intensity; (3) GSACS Horvitz-Thompson variance estimator is design-unbiased for a single SYS sample; and (4) rules-of-thumb summarized with respect to sampling design and spatial effect improve precision. Because inventorying a mini domain is analogous to inventorying a rare variable, alternative network sampling procedures are also readily available for inventorying cross-classes domains. ⁃ Generalized systematic adaptive cluster sampling devised for inventorying cross-classes domains ⁃ Unbiased Hansen-Hurwitz and Horvitz-Thompson estimators derived for domain estimation ⁃ Analytical and empirical comparisons made within GSACS and with systematic sampling ⁃ GSACS provides estimates identical to SS using observations fewer on the order of 20%–40% ⁃ Network sampling procedures including GSACS are efficient for inventorying small domains

An Efficient Plane-Waves Superposition Method for Improved Spatial Correlation in Simulated Reverberation Chambers

Accurate Rayleigh propagation environments within an ideal reverberation chamber (RC) are synthesized computationally using a superposition of plane-waves (PWs). Randomness in the electro-magnetic (EM) field distribution is achieved by synthesizing many field realizations featuring slant-polarized PWs with randomly defined transverse field components. These PWs propagate along a large number of predefined and uniformly distributed directions determined only once via an efficient spiraling sampling scheme over the unit sphere. Comparisons with theoretical Rayleigh field statistics indicate that the proposed method yields random EM field distributions that in several cases follow more closely the field spatial correlation properties, for comparable numerical burden, than other available numerical techniques. Finally, a formula relating the number of PWs to the size of a desired RC analysis volume was provided.

Common Neighbors Matter: Fast Random Walk Sampling with Common Neighbor Awareness

Random walk is widely applied to sample large-scale graphs due to its simplicity of implementation and solid theoretical foundations of bias analysis. However, its computational efficiency is heavily limited by the slow convergence rate (a.k.a. long burn-in period). To address this issue, we propose a common neighbor aware random walk framework called CNARW, which leverages weighted walking by differentiating the next-hop candidate nodes to speed up the convergence. Specifically, CNARW takes into consideration the common neighbors between previously visited nodes and next-hop candidate nodes in each walking step. Based on CNARW, we further develop two efficient unbiased sampling schemes, and we also design two variant algorithms which can reduce sampling cost and speed up the convergence. Experimental results on real-world network datasets show that our approach converges remarkably faster than the state-of-the-art random walk sampling algorithms; and to achieve the same estimation accuracy, our approach reduces the query cost significantly. Last, we use two case studies to demonstrate the effectiveness of our sampling framework in solving large-scale graph analysis tasks.

Quasi-Monte Carlo Approximations for Exponentiated Quadratic Kernel in Latent Force Models

In this project, we consider obtaining Fourier features via more efficient sampling schemes to approximate the kernel in LFMs. A latent force model (LFM) is a Gaussian process whose covariance functions follow an Exponentiated Quadratic (EQ) form, and the solutions for the cross-covariance are expensive due to the computational complexity. To reduce the complexity of mathematical expressions, random Fourier features (RFF) are applied to approximate the EQ kernel. Usually, the random Fourier features are implemented with Monte Carlo sampling, but this project proposes replacing the Monte-Carlo method with the Quasi-Monte Carlo (QMC) method. The first-order and second-order models’ experiment results demonstrate the decrease in NLPD and NMSE, which revealed that the models with QMC approximation have better performance.

An improved fuzzy adaptive teaching learning-based optimisation algorithm for generating pairwise test suites

Pairwise testing has proved its applicability to adequately test software with a huge number of inputs. It can avoid the otherwise impractical exhaustive testing by employing an efficient sampling strategy. Strategies based on meta-heuristic algorithms offer optimal pairwise test suite sizes for software applications. A fuzzy adaptive teaching learning-based optimisation (ATLBO) algorithm has shown competitiveness against other meta-heuristic-based strategies in terms of pairwise test suite generation. Although useful, the present design of ATLBO is lacking in dealing with stagnation or abnormal convergence after some iterations. A remedial operator is introduced in ATLBO in order to address this issue and hence further enhance its convergence speed. With this modification, ATLBO is used for the pairwise test suite generation problem. From the conducted experiments, it can be concluded that the performance of ATLBO with remedial operator is comparable with other pairwise strategies based on hyper-heuristic, meta-heuristic and greedy algorithms.

Investigation of determinism-related issues in the Sobol′ low-discrepancy sequence for producing sound global sensitivity analysis indices

A computationally efficient and robust sampling scheme can support a sensitivity analysis of models to discover their behaviour through Quasi Monte Carlo approximation. This is especially useful for complex models, as often occur in environmental domains when model runtime can be prohibitive. The Sobol' sequence is one of the most used quasi-random low-discrepancy sequences as it can explore the parameter space significantly more evenly than pseudo-random sequences. The built-in determinism of the Sobol' sequence assists in achieving this attractive property. However, the Sobol' sequence tends to deteriorate in the sense that the estimated errors are distributed inconsistently across model parameters as the dimensions of a model increase. By testing multiple Sobol' sequence implementations, it is clear that the deterministic nature of the Sobol' sequence occasionally introduces relatively large errors in sensitivity indices produced by well-known global sensitivity analysis methods, and that the errors do not diminish by averaging through multiple replications. Problematic sensitivity indices may mistakenly guide modellers to make type I and II errors in trying to identify sensitive parameters, and this will potentially impact model reduction attempts based on these sensitivity measurements. This work investigates the cause of the Sobol' sequence's determinism-related issues.

Generalization of the power‐law rating curve using hydrodynamic theory and Bayesian hierarchical modeling

AbstractThe power‐law rating curve has been used extensively in hydraulic practice and hydrology. It is given by , where is discharge, is water elevation, , , and are unknown parameters. We propose a novel extension of the power‐law rating curve, referred to as the generalized power‐law rating curve. It is constructed by linking the physics of open channel flow to a model of the form . The function is referred to as the power‐law exponent and it depends on the water elevation. The proposed model and the power‐law model are fitted within the framework of Bayesian hierarchical models. By exploring the properties of the proposed rating curve and its power‐law exponent, we find that cross‐sectional shapes that are likely to be found in nature are such that the power‐law exponent will usually be in the interval . This fact is utilized for the construction of prior densities for the model parameters. An efficient Markov chain Monte Carlo sampling scheme, that utilizes the lognormal distributional assumption at the data level and Gaussian assumption at the latent level, is proposed for the two models. The two statistical models were applied to four datasets. In the case of three datasets the generalized power‐law rating curve gave a better fit than the power‐law rating curve while in the fourth case the two models fitted equally well and the generalized power‐law rating curve mimicked the power‐law rating curve. We developed an R package, bdrc, for fitting the power‐law and generalized power‐law rating curve models. It is available on CRAN.

How Can Nursing Teams Respond to Large-Scale COVID-19 Screening?

The COVID-19 virus has devastated lives and economies worldwide. The responses of nursing teams to large-scale COVID-19 screening have rarely been addressed or described. The aim of this study is to introduce an efficient response strategy for nurses in large-scale COVID-19 screening. A new COVID-19 case was confirmed on Jan 14, 2021 in Nanning, China. Immediately, a large-scale COVID-19 screening was launched and ran from Jan 14 to Jan 17, 2021. Our nurse team responding to the screening included three major components: (1) establishing a leadership group and a nucleic acid sampling emergency team; (2) defining, conducting, and evaluating nurse training; (3) implementing efficient sampling schemes (10 in 1 mixed sample technique). A total of 500 nurse volunteers were recruited and divided into three echelons. A total of 353 trained nurses were sent to 65 sampling stand stations. In cooperation with nurses from other health institutions, samples were collected from a total of 854,215 people in only 4 days for 2019-nCOV nucleic acid screening. The preparation and efficient response strategies used to conduct this screening may provide a baseline reference for future large-scale COVID-19 screening worldwide.

Towards computationally efficient prediction of molecular signatures from routine histology images

Towards computationally efficient prediction of molecular signatures from routine histology images

Reliability-based topology optimization using stochastic gradients

This paper addresses the computational challenges in reliability-based topology optimization (RBTO) of structures associated with the estimation of statistics of the objective and constraints using standard sampling methods. The aim is to overcome the accuracy issues of traditional methods that rely on approximating the limit-state function. Herein, we present a stochastic gradient-based approach, where we estimate the probability of failure at every few optimization iterations using an efficient sampling strategy. To estimate the gradients of the failure probability with respect to the design parameters, we apply Bayes’ rule wherein we assume a parametric exponential model for the probability density function of the design parameters conditioned on the failure. The design parameters and the parameters of this probability density function are updated using a stochastic gradient descent approach requiring only a small, e.g., $${\mathcal {O}}(1)$$ , number of random samples per iteration, thus, leading to considerable reduction of the computational cost as compared to standard RBTO techniques. We illustrate the proposed approach with a benchmark example that has an analytical solution as well as two widely used problems in structural topology optimization. These examples illustrate the efficacy of the approach in producing reliable designs.

Estimating viral prevalence with data fusion for adaptive two-phase pooled sampling.

The COVID‐19 pandemic has highlighted the importance of efficient sampling strategies and statistical methods for monitoring infection prevalence, both in humans and in reservoir hosts. Pooled testing can be an efficient tool for learning pathogen prevalence in a population. Typically, pooled testing requires a second‐phase retesting procedure to identify infected individuals, but when the goal is solely to learn prevalence in a population, such as a reservoir host, there are more efficient methods for allocating the second‐phase samples.To estimate pathogen prevalence in a population, this manuscript presents an approach for data fusion with two‐phased testing of pooled samples that allows more efficient estimation of prevalence with less samples than traditional methods. The first phase uses pooled samples to estimate the population prevalence and inform efficient strategies for the second phase. To combine information from both phases, we introduce a Bayesian data fusion procedure that combines pooled samples with individual samples for joint inferences about the population prevalence.Data fusion procedures result in more efficient estimation of prevalence than traditional procedures that only use individual samples or a single phase of pooled sampling.The manuscript presents guidance on implementing the first‐phase and second‐phase sampling plans using data fusion. Such methods can be used to assess the risk of pathogen spillover from reservoir hosts to humans, or to track pathogens such as SARS‐CoV‐2 in populations.

Sector subsampling for basal area ratio estimation: an alternative to big BAF sampling

Big basal area factor (big BAF) sampling is a widely used subsampling method to select measure-trees. Several studies have shown big BAF sampling to be an efficient sampling scheme. In this study, we use sector sampling (Smith et al. 2008, For. Sci. 54: 67–76) as an alternative subsample selection method. Based on some simulated mapped stands derived from three balsam fir (Abies balsamea (L.) Mill.) spacing trials in western Newfoundland, we show that sector subsampling is comparable to big BAF sampling in terms of estimated mean basal area ratios and their associated standard errors. Differences between big BAF sampling and sector sampling methods showed less than 1% difference across the three sites. As with big BAF sampling, changes in sample intensity had no significant (p < 0.05) effects on the accuracy of estimating mean biomass to basal area ratios and the resulting estimated mean biomasses per unit area.

A Dynamic Spatial Model for Chronic Wasting Disease in Colorado

A spatio-temporal statistical model for Chronic Wasting Disease is presented. The model has underpinnings from traditional epidemic models with differential equations and uses a Bayesian hierarchy to directly incorporate existing prevalence data. Spatial dynamics are modeled explicitly through a system of difference equations rather than through covariance. The posterior distribution gives evidence of a long term stable level of disease prevalence, and approximates the probability of the movement of the disease from one area to another. Predictions for the future of Chronic Wasting Disease in Colorado are given. The model is used to formulate efficient sampling schemes for future data collection.

Variational Monte Carlo Calculations of A≤4 Nuclei with an Artificial Neural-Network Correlator Ansatz.

The complexity of many-body quantum wave functions is a central aspect of several fields of physics and chemistry where nonperturbative interactions are prominent. Artificial neural networks (ANNs) have proven to be a flexible tool to approximate quantum many-body states in condensed matter and chemistry problems. In this work we introduce a neural-network quantum state ansatz to model the ground-state wave function of light nuclei, and approximately solve the nuclear many-body Schrödinger equation. Using efficient stochastic sampling and optimization schemes, our approach extends pioneering applications of ANNs in the field, which present exponentially scaling algorithmic complexity. We compute the binding energies and point-nucleon densities of A≤4 nuclei as emerging from a leading-order pionless effective field theory Hamiltonian. We successfully benchmark the ANN wave function against more conventional parametrizations based on two- and three-body Jastrow functions, and virtually exact Green's function MonteCarlo results.

Optimal sampling of dynamical large deviations via matrix product states.

The large deviation statistics of dynamical observables is encoded in the spectral properties of deformed Markov generators. Recent works have shown that tensor network methods are well suited to compute accurately the relevant leading eigenvalues and eigenvectors. However, the efficient generation of the corresponding rare trajectories is a harder task. Here, we show how to exploit the matrix product state approximation of the dominant eigenvector to implement an efficient sampling scheme which closely resembles the optimal (so-called "Doob") dynamics that realizes the rare events. We demonstrate our approach on three well-studied lattice models, the Fredrickson-Andersen and East kinetically constrained models, and the symmetric simple exclusion process. We discuss how to generalize our approach to higher dimensions.

Quantifying soil carbon in temperate peatlands using a mid-IR soil spectral library

Abstract. Traditional laboratory methods for acquiring soil information remain important for assessing key soil properties, soil functions and ecosystem services over space and time. Infrared spectroscopic modeling can link and massively scale up these methods for many soil characteristics in a cost-effective and timely manner. In Switzerland, only 10 % to 15 % of agricultural soils have been mapped sufficiently to serve spatial decision support systems, presenting an urgent need for rapid quantitative soil characterization. The current Swiss soil spectral library (SSL; n = 4374) in the mid-infrared range includes soil samples from the Biodiversity Monitoring Program (BDM), arranged in a regularly spaced grid across Switzerland, and temporally resolved data from the Swiss Soil Monitoring Network (NABO). Given that less than 2 % of the samples in the SSL originate from organic soils, we aimed to develop both an efficient calibration sampling scheme and accurate modeling strategy to estimate the soil carbon (SC) contents of heterogeneous samples between 0 and 2 m depth from 26 locations within two drained peatland regions (School of Agricultural, Forest and Food Sciences (HAFL) data set; n = 116). The focus was on minimizing the need for new reference analyses by efficiently mining the spectral information of the SSL. We used partial least square regressions (PLSRs), together with five repetitions of a location-grouped, 10-fold cross-validation, to predict SC ranging from 1 % to 52 % in the local HAFL data set. We compared the validation performance of different calibration schemes involving local models (1), models using the entire SSL combined with local samples (2), commonly referred to as spiking, and subsets of local and SSL samples optimized for the peatland target sites using the resampling local (RS-LOCAL) algorithm (3). Using local and RS-LOCAL calibrations with at least five local samples, we achieved similar validation results for predictions of SC up to 52 % (R2 = 0.93 to 0.97; bias = -0.07 to 1.65; root mean square error (RMSE) = 2.71 % to 3.89 % total carbon; ratio of performance to deviation (RPD) = 3.38 to 4.86; and ratio of performance to interquartile range (RPIQ) = 4.93 to 7.09). However, calibrations using RS-LOCAL only required five or 10 local samples for very accurate models (RMSE = 3.16 % and 2.71 % total carbon, respectively), while purely local calibrations required 50 samples for similarly accurate results (RMSE < 3 % total carbon). Of the three approaches, the entire SSL spiked with local samples for model calibration led to validations with the lowest performance in terms of R2, bias, RMSE, RPD and RPIQ. Hence, we show that a simple and comprehensible modeling approach, using RS-LOCAL together with a SSL, is an efficient and accurate strategy when using infrared spectroscopy. It decreases field and laboratory work, the bias of SSL spiking approaches and the uncertainty of local models. If adequately mined, the information in the SSL is sufficient to predict SC in new and independent study regions, even if the local soil characteristics are very different from the ones in the SSL. This will help to efficiently scale up the acquisition of quantitative soil information over space and time.

The Modified-Half-Normal distribution: Properties and an efficient sampling scheme

We introduce a new family of probability distributions that we call the Modified-Half-Normal distributions. It is supported on the positive part of the real line with its probability density proportional to the function We explore a number of its properties including showing the fact that the normalizing constant and moments of the distribution can be represented in terms of the Fox-Wright function. We demonstrate its relevance by showing its connection to a number of Bayesian statistical methods appearing from multiple areas of research such as Bayesian Binary regression, Analysis of Directional data, and Bayesian graphical model. The availability of its efficient sampling scheme is important to the success of such Bayesian procedures. Therefore, a major focus of this article is the development of methods for generating random samples from the Modified-Half-Normal distribution. To ensure efficiency, we prove that the constructed accept reject algorithms are “uniformly efficient” with high acceptance probability irrespective of the choice of the parameter specifications. Finally, we provide basic inference procedure for its parameters when analyzing data assuming the Modified-Half-Normal probability model.

To Sobol or not to Sobol? The effects of sampling schemes in systems biology applications

Computational and mathematical models in biology rely heavily on the parameters that characterize them. However, robust estimates for their values are typically elusive and thus a large parameter space becomes necessary for model study, particularly to make translationally impactful predictions. Sampling schemes exploring parameter spaces for models are used for a variety of purposes in systems biology, including model calibration and sensitivity analysis. Typically, random sampling is used; however, when models have a high number of unknown parameters or the models are highly complex, computational cost becomes an important factor. This issue can be reduced through the use of efficient sampling schemes such as Latin hypercube sampling (LHS) and Sobol sequences. In this work, we compare and contrast three sampling schemes – random sampling, LHS, and Sobol sequences – for the purposes of performing both parameter sensitivity analysis and model calibration. In addition, we apply these analyses to different types of computational and mathematical models of varying complexity: a simple ODE model, a complex ODE model, and an agent-based model. In general, the sampling scheme had little effect when used for calibration efforts, but when applied to sensitivity analyses, Sobol sequences exhibited faster convergence. While the observed benefit to convergence is relatively small, Sobol sequences are computationally less expensive to compute than LHS samples and also have the benefit of being deterministic, which allows for better reproducibility of results.