Large Sample Sets Research Articles

Prediction of wheat quality parameters combining Raman, fluorescence, and near‐infrared spectroscopy (NIRS)

AbstractBackground and ObjectivesA high amount of the world wheat production is used for bread making. Due to its simplicity, wheat samples are traded based on protein content, although this is only an indirect measurement of baking quality. As direct bread quality traits like loaf volume, water absorption, and dough properties are cumbersome to measure along wheat supply chains and in wheat breeding programs, this study aims to predict directly dough properties and baking behavior by combining complementary information from near‐infrared, Raman, and fluorescence spectroscopy. Several proven data preprocessing and regression algorithms were evaluated, including partial least‐squares regression (PLSR) and genetic algorithm (GA).FindingsA highly diverse sample set of 415 wheat samples from a wheat hybrid breeding program was analyzed as whole grain, whole grain flour, and extracted flour. The best predictive models yielded a 10‐fold cross‐validated coefficient of determination R2 of .973, .873, .774, .835, .369, .447, .311, .512, .536, .723, and .715 for protein content, wet gluten, loaf volume, water uptake, elongation resistance to extension, extensibility, ratio number, extensograph energy, maltose number, plant height, and grain yield, respectively.ConclusionThe prediction accuracies are well suited for screening purposes in wheat breeding and allow better quality assessment compared to protein content along wheat supply chains.Significance and NoveltyIn this study, complementary spectroscopic techniques are combined for the first time to achieve higher prediction accuracy for a wide range of wheat quality parameters. In addition, Raman and fluorescence spectroscopy were used to predict quality data from a large sample set.

A Stochastic Discrete Empirical Interpolation Approach for Parameterized Systems

As efficient separation of variables plays a central role in model reduction for nonlinear and nonaffine parameterized systems, we propose a stochastic discrete empirical interpolation method (SDEIM) for this purpose. In our SDEIM, candidate basis functions are generated through a random sampling procedure, and the dimension of the approximation space is systematically determined by a probability threshold. This random sampling procedure avoids large candidate sample sets for high-dimensional parameters, and the probability based stopping criterion can efficiently control the dimension of the approximation space. Numerical experiments are conducted to demonstrate the computational efficiency of SDEIM, which include separation of variables for general nonlinear functions, e.g., exponential functions of the Karhu nen–Loève (KL) expansion, and constructing reduced order models for FitzHugh–Nagumo equations, where symmetry among limit cycles is well captured by SDEIM.

Comparing Boron Soil Testing Methods for Coastal Plain Sandy Soils

ABSTRACT There are several methods for determining boron (B) in soil for soil testing purposes, yet variability among methods can lead to inconsistencies in determined concentrations for the same soil sample. The most typical method used to determine B in soil is hot-water B (HWB) extraction, but this method has reduced efficiency when dealing with large sets of soil samples along with other disadvantages such as non-uniformity in the boiling of samples. As a result, universal extractants have become increasingly popular for the extraction of B due to enhanced efficiency in a high throughput laboratory setting. This study uses a Deming regression model to compare the traditional HWB extraction method with two common universal extractants, Mehlich-1 (M1), and Mehlich-3 (M3), for coastal plain sandy soils. Results showed M3-B extractability is consistently higher than M1-B and HWB for all tested soils. The M3 method was then verified for precision using blind extension soil samples. The hot-water method was unsuccessful in producing reliable extractable B concentration in soils with high calcium carbonate and organic matter contents. Results from this study can be used to optimize B determination for high throughput production laboratory settings.

Physics-Based Simulation and Machine Learning for the Practical Implementation of EIT-Based Tactile Sensors

This work reports an innovative framework toward the practical implementation of tactile sensors based on electrical impedance tomography. Particularly, we tested our framework on the development of a piezoresistive insole-shaped sensor prototype. Our approach couples the implementation of the forward model through a physics-based general purpose FEM software and an ANN model for the inverse problem solution. First, we developed a FEM forward model in COMSOL Multyphysics, and we optimized the parameters of the model to better resemble the prototype characteristics. Then, we trained an ANN model with an “artificial” dataset generated by feeding the forward model with a large set of different conductivity distribution samples and measuring the voltage at the boundary electrodes. For comparison, we employed the forward model also to compute the sensitivity matrix, which is required to apply standard linear reconstruction methods. We statistically compared the performance of the proposed machine learning approach with those of standard linear reconstruction methods. The results on simulated data highlight the higher accuracy of the ANN with respect to the other methods. In particular, the mean conductivity RMSE is 0.8 S/m. Finally, we tested our approach with the physical prototype to evaluate the performance in touch position detection. Again, the ANN achieves the minimum mean position error (5.74 mm), demonstrating the feasibility of using machine learning trained with artificial datasets for solving the inverse problem in EIT-based tactile sensors.

The Mixed-Body Model: A Method for Predicting Large Deflections in Stepped Cantilever Beams

AbstractThis paper presents a method for predicting endpoint coordinates, stress, and force to deflect stepped cantilever beams under large deflections. This method, the mixed-body model or MBM, combines small deflection theory and the pseudo-rigid-body model for large deflections. To analyze the efficacy of the model, the MBM is compared to a model that assumes the first step in the beam to be rigid, to finite element analysis, and to the numerical boundary value solution over a large sample set of loading conditions, geometries, and material properties. The model was also compared to physical prototypes. In all cases, the MBM agrees well with expected values. Optimization of the MBM parameters yielded increased agreement, leading to average errors of <0.01 to 3%. The model provides a simple, quick solution with minimal error that can be particularly helpful in design.

An Online Kernel Adaptive Filtering-Based Approach for Mid-Price Prediction

The idea of multivariate and online stock price prediction via the kernel adaptive filtering (KAF) paradigm is proposed in this article. The prediction of stock prices is traditionally done with regression and classification, thereby requiring a large set of batch-oriented and independent training samples. This is problematic considering the nonstationary nature of a financial time series. In this research, we propose an online kernel adaptive filtering-based approach for stock price prediction to overcome this challenge. To examine a stock's performance and demonstrate the work's superiority, we use ten different KAF family of algorithms. In this paper, we take on this challenge and propose an approach for predicting stock prices. To analyze a stock's performance and demonstrate the work's superiority, we use ten distinct KAF algorithms. Besides, the results are analyzed on nine-time windows such as one day, sixty minutes, thirty minutes, twenty five minutes, twenty minutes, fifteen minutes, ten minutes, five minutes, and one minute. We are the first to experiment with several time windows for all fifty stocks on the Indian National Stock Exchange, to the best of our knowledge. It should be noted here that the experiments are performed on stocks making up the main index: Nifty-50. In terms of performance and compared to existing methods, we have a 66% probability of correctly predicting a stock's next upward or downward movement. This number clearly shows the edge that the proposed method has in actual deployment. Furthermore, the experimental findings show that KAF is not only a better option for predicting stock prices but that it may also be used as an alternative in high-frequency trading due to its low latency.

Sensing of Soil Organic Matter Using Laser-Induced Breakdown Spectroscopy Coupled with Optimized Self-Adaptive Calibration Strategy.

Rapid quantification of soil organic matter (SOM) is a great challenge for the health assessment and fertility management of agricultural soil. Laser-induced breakdown spectroscopy (LIBS) with appropriate modeling algorithms is an alternative tool for this measurement. However, the current calibration strategy limits the prediction performance of the LIBS technique. In this study, 563 soil samples from Hetao Irrigation District in China were collected; the LIBS spectra of the soils were recorded in the wavenumber range of 288–950 nm with a resolution of 0.116 nm; a self-adaptive partial least squares regression model (SAM–PLSR) was employed to explore optimal model parameters for SOM prediction; and calibration parameters including sample selection for the calibration database, sample numbers and sample location sites were optimized. The results showed that the sample capacity around 60–80, rather than all of the samples in the soil library database, was selected for calibration from a spectral similarity re-ordered database regarding unknown samples; the model produced excellent predictions, with R2 = 0.92, RPD = 3.53 and RMSEP = 1.03 g kg−1. Both the soil variances of the target property and the spectra similarity of the soil background were the key factors for the calibration model, and the small sample set led to poor predictions due to the low variances of the target property, while negative effects were observed for the large sample set due to strong interferences from the soil background. Therefore, the specific unknown sample depended strategy, i.e., self-adaptive modelling, could be applied for fast SOM sensing using LIBS for soils in varied scales with improved robustness and accuracy.

What can we learn from amino acids about oceanic organic matter cycling and degradation?

Abstract. Amino acids (AAs) mainly bound in proteins are major constituents of living biomass and non-living organic material in the oceanic particulate and dissolved organic matter pool. Uptake and cycling by heterotrophic organisms lead to characteristic changes in AA composition so that AA-based biogeochemical indicators are often used to elucidate processes of organic matter cycling and degradation. We analyzed particulate AA in a large sample set collected in various oceanic regions covering sinking and suspended particles in the water column, sediment samples, and dissolved AA from water column and pore water samples. The aim of this study was to test and improve the use of AA-derived biogeochemical indicators as proxies for organic matter sources and degradation and to better understand particle dynamics and interaction between the dissolved and particulate organic matter pools. A principal component analysis (PCA) of all data delineates diverging AA compositions of sinking and suspended particles with increasing water depth. A new sinking particle and sediment degradation indicator (SDI) allows a fine-tuned classification of sinking particles and sediments with respect to the intensity of degradation, which is associated with changes of stable isotopic ratios of nitrogen (δ15N). This new indicator is furthermore sensitive to sedimentary redox conditions and can be used to detect past anoxic early diagenesis. A second indicator emerges from the AA spectra of suspended particulate matter (SPM) in the epipelagic and that of the meso- and bathypelagic ocean and is a residence time indicator (RTI). The characteristic changes in AA patterns from shallow to deep SPM are recapitulated in the AA spectra of the dissolved organic matter (DOM) pool, so that deep SPM is more similar to DOM than to any of the other organic matter pools. This implies that there is equilibration between finely dispersed SPM and DOM in the deep sea, which may be driven by microbial activity combined with annealing and fragmentation of gels. As these processes strongly depend on physico-chemical conditions in the deep ocean, changes in quality and degradability of DOM may strongly affect the relatively large pool of suspended and dissolved AA in the ocean that amounts to 15 Pg amino acid carbon (AAC) and 89 ± 29 Pg AAC, respectively.

DirectDetect SARS-CoV-2 Direct Real-Time RT-PCR Study Using Patient Samples.

COVID-19 is an infectious disease that caused a global pandemic affecting people worldwide. As disease detection and vaccine rollout continue to progress, there is still a need for efficient diagnostic tools to satisfy continued testing needs. This preliminary study evaluated a novel SARS-CoV-2 diagnostic test called DirectDetect SARS-CoV-2 Direct Real-time reverse transcriptase polymerase chain reaction (RT-PCR) based on a limited sample size of 24 respiratory samples from 14 SARS-CoV-2-positive patients. The test is advantageous compared to others on the market since it does not require viral transport medium or viral RNA extraction prior to nucleic acid amplification and detection. This capability transforms the hours-long sample preparation time into a minutes-long procedure while also eliminating the need for many costly reagents which may be difficult to obtain during the surge in nucleic acid-based testing during the pandemic. The results show a positive agreement of 94.7, 100, and 94.7% between dry sample swabs, treated samples, and untreated samples tested using the DirectDetect SARS-CoV-2 Direct Real-time RT-PCR compared to tests used in a clinical laboratory, respectively. The findings indicate that DirectDetect can be used for multiple different sample types while reducing the number of reagents and time needed for diagnosis. Although this study shows promising results using the DirectDetect results, further validation of this test using a larger sample set is required to assess the true performance of this test.

Evaluation of SARS-CoV-2 rapid antigen diagnostic tests for saliva samples

Using saliva samples would facilitate sample collection, diagnostic feasibility, and mass screening of SARS-CoV-2. We tested two rapid antigen (RAD) immunochromatographic tests designed for detection of SARS-CoV-2 in saliva: Rapid Response™ COVID-19 Antigen Rapid Test Cassette for oral fluids and DIAGNOS™ COVID-19 Antigen Saliva Test. Evaluation of detection limit was performed with purified SARS-CoV-2 nucleocapsid protein and live SARS-CoV-2 virus. Sensitivity and specificity were further evaluated with reverse transcription quantitative PCR (RT-qPCR) positive and negative saliva samples from hospitalized individuals with COVID-19 (n = 39) and healthcare workers (n = 20). DIAGNOS showed higher sensitivity than Rapid Response for both nucleocapsid protein and live virus. The limit of detection of the saliva test from DIAGNOS was further comparable with the Abbott Panbio™ COVID-19 Ag Rapid Test designed for nasopharyngeal samples. DIAGNOS and Rapid Response detected nine (50.0%) and seven (38.9%), respectively, of the 18 RT-qPCR positive saliva samples. All RT-qPCR negative saliva (n = 41) were negative with both tests. Only one of the RT-qPCR positive saliva samples contained infectious virus as determined by cell culture and was also positive using the saliva RADs. The results show that the DIAGNOS may be an important and easy-to-use saliva RAD complement to detect SARS-CoV-2 positive individuals, but validation with a larger sample set is warranted.

Analysis of Image Evolution of Ancient Large Figurines Based on Deep Neural Network

Unlike traditional image recognition technology, DL can automatically extract features and improve recognition accuracy by combining feature extraction and classification. The challenges and shortcomings of traditional image recognition methods are discussed in this article, as well as the development process and research status of DL. Related theories in image recognition based on deep learning (DL) are proposed, DL’s basic models and methods are analyzed, and related image data sets are demonstrated experimentally. Furthermore, because DL is typically used for large sample sets, this paper proposes an improved algorithm based on small samples, as well as a DNN-based analysis model for the evolution of ancient large figurine images. This model, when compared to the traditional neural network model, can speed up the network’s convergence speed and reduce training time to a certain extent. This model improves the rate of image recognition while lowering the error rate.

Simple and Robust Detection of CYP2D6 Gene Deletions and Duplications Using CYP2D8P as Reference.

Genotyping of the CYP2D6 gene is the most commonly applied pharmacogenetic test globally. Significant economic interests have led to the development of a plurality of assays, available for almost any genotyping platform or DNA detection chemistry. Of all the genetic variants, copy number variations are particular difficult to detect by polymerase chain reaction. Here, we present two simple novel approaches for the identification of samples carrying either deletions or duplications of the CYP2D6 gene; by relative quantification using a singleplex 5′nuclease real-time PCR assay, and by high-resolution melting of PCR products. These methods make use of universal primers, targeting both the CYP2D6 and the reference gene CYP2D8P, which is necessary for the analysis. The assays were validated against a reference method using a large set of samples. The singleplex nature of the 5′nuclease real-time PCR ensures that the primers anneal with equal affinity to both the sequence of the CYP2D6 and the reference gene. This facilitates robust identification of gene deletions and duplications based on the cycle threshold value. In contrast, the high-resolution melting assay is an end-point PCR, where the identification relies on variations between the amount of product generated from each of the two genes.

Performance enhancement‐based active learning sample selection method

AbstractRepresentative samples are important for multivariate calibration. The highly efficient selection of representative samples to be labelled can save money and time. Existing methods, such as Kennard‐Stone and net analyte signal selection, are usually based on the distance between candidate samples and labelled calibration sets in feature space. However, these distances are influenced by the feature space, which is spanned by an information vector extracted from labelled samples. To overcome the negative effects of the distance‐based selection method, a model performance enhancement‐based sample selection method is proposed to select calibration samples efficiently. Based on loss function optimization, the samples that can improve model performance the most, as estimated by bootstrap, are sequentially selected and added to the calibration set. Due to the high representation of each sample, a few samples can build a model that has no significant loss of prediction ability when compared with a model built with the large number set of calibration samples. The performance enhancement‐based active learning (PEAL) sample selection method is both effective and efficient.

Transformer Neural Network for Weed and Crop Classification of High Resolution UAV Images

Monitoring crops and weeds is a major challenge in agriculture and food production today. Weeds compete directly with crops for moisture, nutrients, and sunlight. They therefore have a significant negative impact on crop yield if not sufficiently controlled. Weed detection and mapping is an essential step in weed control. Many existing research studies recognize the importance of remote sensing systems and machine learning algorithms in weed management. Deep learning approaches have shown good performance in many agriculture-related remote sensing tasks, such as plant classification, disease detection, etc. However, despite the success of these approaches, they still face many challenges such as high computation cost, the need of large labelled datasets, intra-class discrimination (in growing phase weeds and crops share many attributes similarity as color, texture, and shape), etc. This paper aims to show that the attention-based deep network is a promising approach to address the forementioned problems, in the context of weeds and crops recognition with drone system. The specific objective of this study was to investigate visual transformers (ViT) and apply them to plant classification in Unmanned Aerial Vehicles (UAV) images. Data were collected using a high-resolution camera mounted on a UAV, which was deployed in beet, parsley and spinach fields. The acquired data were augmented to build larger dataset, since ViT requires large sample sets for better performance, we also adopted the transfer learning strategy. Experiments were set out to assess the effect of training and validation dataset size, as well as the effect of increasing the test set while reducing the training set. The results show that with a small labeled training dataset, the ViT models outperform state-of-the-art models such as EfficientNet and ResNet. The results of this study are promising and show the potential of ViT to be applied to a wide range of remote sensing image analysis tasks.

Loss of circadian gene Timeless induces EMT and tumor progression in colorectal cancer via Zeb1-dependent mechanism.

The circadian gene Timeless (TIM) provides a molecular bridge between circadian and cell cycle/DNA replication regulatory systems and has been recently involved in human cancer development and progression. However, its functional role in colorectal cancer (CRC), the third leading cause of cancer-related deaths worldwide, has not been fully clarified yet. Here, the analysis of two independent CRC patient cohorts (total 1159 samples) reveals that loss of TIM expression is an unfavorable prognostic factor significantly correlated with advanced tumor stage, metastatic spreading, and microsatellite stability status. Genome-wide expression profiling, in vitro and in vivo experiments, revealed that TIM knockdown induces the activation of the epithelial-to-mesenchymal transition (EMT) program. Accordingly, the analysis of a large set of human samples showed that TIM expression inversely correlated with a previously established gene signature of canonical EMT markers (EMT score), and its ectopic silencing promotes migration, invasion, and acquisition of stem-like phenotype in CRC cells. Mechanistically, we found that loss of TIM expression unleashes ZEB1 expression that in turn drives the EMT program and enhances the aggressive behavior of CRC cells. Besides, the deranged TIM-ZEB1 axis sets off the accumulation of DNA damage and delays DNA damage recovery. Furthermore, we show that the aggressive and genetically unstable 'CMS4 colorectal cancer molecular subtype' is characterized by a lower expression of TIM and that patients with the combination of low-TIM/high-ZEB1 expression have a poorer outcome. In conclusion, our results as a whole suggest the engagement of an unedited TIM-ZEB1 axis in key pathological processes driving malignant phenotype acquisition in colorectal carcinogenesis. Thus, TIM-ZEB1 expression profiling could provide a robust prognostic biomarker in CRC patients, supporting targeted therapeutic strategies with better treatment selection and patients' outcomes.

Linkage, recombination and mutation rate analyses of 16 X-chromosomal STR loci in Sri Lankan Sinhalese pedigrees.

This study investigated genetic linkage, recombination fractions and mutation rates of 16 X chromosomal short tandem repeat (X-STR) markers using a recently developed multiplex PCR assay for Sinhalese population of Sri Lanka, by analyzing 81 three-generation families including 81 grandfathers with daughters and 162 grandsons. In addition, 31 two generation families involving mother father daughter trios were included for mutation analysis. The analysis of recombination fractions between marker pairs identified two linkage blocks (maximum LOD scores > 3.0) each spanning a physical distance of 44.35Mb and 6.04Mb respectively. Though recombination events are usually rare among closely linked markers, crossovers were observed for markers located < 1.0Mb apart. The recombination fractions observed are not fully concordant with those reported earlier, including the second-generation Rutgers combined linkage-physical map. This suggests that linkage is not uniform among different populations. However, the overall and marker-specific mutation rates of the present study did not differ from previous reports, though it needs confirmation with a much larger sample set. The findings presented here will provide the baseline information required for biostatistical calculations conducted using X-STR markers, in complex kinship analysis of Sinhalese population.

Analysis of the Urbach tail in cesium lead halide perovskites

The role of structural and dynamical disorder in semiconductors is a topic of fundamental relevance because of its contribution to the spectral line shape of the photoluminescence, and it plays a major role in ruling the carrier transport properties at the band edge. In this regard, a class of semiconductors, i.e., halide perovskites, deeply investigated in the last decade, shows a peculiar degree of disorder, which has only been recently under investigation. The interest to study disorder in halide perovskites is related to the large set of innovative applications of this class of materials, spanning from energy harvesting to high brilliance incoherent and coherent light emitters. In this perspective, we show that quantitative information on the disorder in halide perovskites can be extracted by deep analysis of the photoluminescence in different experimental conditions. Our study, conducted on a large set of samples of a metal halide perovskite, CsPbBr3, prepared with various synthesis/deposition methods, clarifies the relative weight of the static and dynamic contributions. A comparison with theoretical predictions is provided, gaining insights into the exciton/carrier–phonon interaction in metal halide perovskites.

Improvement and Evaluation of Data Consistency Metric CIL for Software Engineering Data Sets

Software data sets derived from actual software products and their development processes are widely used for project planning, management, quality assurance and process improvement, etc. Although it is demonstrated that certain data sets are not fit for these purposes, the data quality of data sets is often not assessed before using them. The principal reason for this is that there are not many metrics quantifying fitness of software development data. In that respect, this study makes an effort to fill in the void in literature by devising a new and efficient assessment method of data quality. To that end, we start as a reference from Case Inconsistency Level (CIL), which counts the number of inconsistent project pairs in a data set to evaluate its consistency. Based on a follow-up evaluation with a large sample set, we depict that CIL is not effective in evaluating the quality of certain data sets. By studying the problems associated with CIL and eliminating them, we propose an improved metric called Similar Case Inconsistency Level (SCIL). Our empirical evaluation with 54 data samples derived from six large project data sets show that SCIL can distinguish between consistent and inconsistent data sets, and that prediction models for software development effort and productivity built from consistent data sets achieve indeed a relatively higher accuracy.

The production of protective earth-based mortars for earth constructions in southeastern Brazil during the 19th century coffee economy

ABSTRACT The study of materials used in ancient buildings provides a means to shed light on traditional building practices of past societies, but can serve as inspiration to tackle challenges faced by the current generation. Characterising earth-based structures and finishing elements is part of this mission, especially at a time when earthen architecture is the subject of renewed interest due to its many advantages, particularly in terms of sustainability. This paper considers a set of historical earthen houses built during the first phase of the coffee economy (1820-1880) in the middle Paraíba do Sul River valley, in southeastern Brazil. Physical (colour and texture) and chemical (FTIR, TGA, XRF, and XRD) analyses performed on a large set of mortar and local soil samples collected in the region formed the basis for discussing possible soil selection criteria as raw materials, texture solutions, and the stabilisation strategy of the final product. This last aspect implied the addition of small quantities of lime, in the case of external renders and more sporadically in plasters, without the use of fibres or organic additives. Chemical data and historical sources suggest that the raw materials used for this purpose were probably brought to the region from coastal areas.

Land Use and Land Cover Area Estimates From Class Membership Probability of a Random Forest Classification

Estimates of the area of land cover classes or land change are frequently calculated from land cover classification maps by counting the pixels labeled as each class in the map. This procedure is known to produce biased estimates of area for many widely used classification algorithms, including random forests. Poststratification estimation using the mapped classes as strata has been proposed to obtain unbiased estimates of the class areas. Still, the method requires additional sampling units, which may not be available or be the most efficient method depending on the application. Alternatively, consistent estimates of class areas can be obtained using class membership probabilities estimates from a random forest classification. This article demonstrates that, for a large sample and proper set of explanatory variables, the error of the predicted class membership probabilities obtained from a random forest classification converges to zero. Therefore, the expected class areas calculated from these probabilities converge to the population class areas. On average, the relative error of the expected class proportions computed by class membership probabilities from a random forests model was 40% points lower than the proportions estimated by pixel counting. Our proposed approach is also comparable to the area-adjusted method, which is currently considered the best practice by the remote sensing community. We recommend that class probability estimates area always retained and used for calculating expected class areas or area proportions based on our results. Our method reduces bias compared to statistics calculated by pixel counting and circumvents the need for poststratification area estimates under certain conditions.