Data Pairs Research Articles

Prediction of the cadmium content in grains of low-accumulating wheat cultivars and soil cadmium threshold for safe production

Wheat cultivars with low cadmium (Cd) accumulation (WCCA-Cd) are widely used on Cd-contaminated farmland and can effectively reduce the grains Cd content to achieve the clean production in agriculture. However, there is a lack of soil Cd content thresholds specifically for the WCCA-Cd. To achieve safer production of wheat grains, a prediction model for WCCA-Cd was developed based on total Cd content in wheat grains (Grain-Cd) and soil factors using 379 soil‒wheat datasets from 10 field experiments in this research. The adaptability of the prediction model was tested using 312 pairs of data. Descriptive statistics (N = 379) showed that soil pH ranged from 4.28 to 8.76, the total Cd content in soil (Soil-Cd) ranged from 0.09 to 10.25 mg kg−1, and Grain-Cd ranged from 0.002 to 1.35 mg kg−1. Correlation analysis showed that Grain-Cd had a highly significant negative correlation with soil organic matter (SOM), soil pH, and soil conductivity (EC) and a highly significant positive correlation with Soil-Cd and Cd content in the diethylenetriaminepentaacetic acid extraction state (DTPA-Cd) (P < 0.001). The bioconcentration factor (BCF) decreased with increaseing soil pH. Both random forest and stepwise regression equations indicated that the most important factors affecting Cd bioeffectiveness were Soil-Cd and soil pH, explaining 80% of the variation in Grain-Cd. The prediction model indicated a soil Cd threshold (0.23–1.25 mg kg−1) for the safe production of WCCA-Cd at soil pH (4–9), which has a wider range of applicability compared with the Chinese Soil Safety Standards. Therefore, these results can quantify the grain Cd content of WCCA-Cd and guide the safe and clean production of WCCA-Cd on different Cd contaminated farmlands.

Validation of the JOYTECH DBP-6279B blood pressure monitor in adults and adolescents according to the AAMI/ESH/ISO universal standard(ISO 81060-2:2018 + Amd.1:2020).

This study aimed to validate the accuracy of DBP-6279B, an automated inflationary oscillometric upper-arm blood pressure (BP) monitor, in the sitting position according to the AAMI/ESH/ISO (81060-2 : 2018 + Amd.1 : 2020) universal standard protocol. SBPs and DBPs were measured simultaneously on the same arm in 88 adults (female : male = 47 : 41) with a mean age of 56.85 years using a mercury sphygmomanometer (two observers) and a DBP-6279B device (one supervisor). The AAMI/ESH/ISO 81060-2 : 2018 and Amd.1 : 2020 universal standards for the validation of BP-measuring devices in adults and adolescents were followed. A total of 259 valid pairs of data were used in the analysis. According to Criterion 1, the mean difference of SBP between the test device (DBP-6279B) and the reference device (the mercury sphygmomanometer) was 0.75 mmHg, with a SD of 7.66 mmHg. The mean difference in DBP was 1.13 mmHg, with a SD of 6.14 mmHg. The mean difference of both SBP and DBP was less than 5 mmHg, and the SD was less than 8 mmHg, which met the requirements. According to Criterion 2, the mean difference of SBP between the test device and the reference device was 0.85 mmHg, and the SD was 6.56 mmHg, which was less than 6.88 mmHg and met the requirements. The mean difference in DBP was 1.27 mmHg, and the SD was 5.42 mmHg, which was less than 6.82 mmHg and met the requirements. DBP-6279B fulfilled the requirements of the AAMI/ESH/ISO universal standard (ISO 81060-2 : 2018 + Amd.1 : 2020); hence, it can be recommended for both clinical and self/home BP measurement in adults and adolescents.

Unpaired sonar image denoising with simultaneous contrastive learning

Implementing underwater sonar image denoising algorithms through unpaired clean and noisy images is feasible and meaningful, since compliant paired real underwater sonar data is nearly impractical. Nevertheless, acquiring high-performance denoising algorithm is challenging without pairs of data as learning objects. In addition, the wide domain gap between noisy and clean samples as well as the noise contained in underwater sonar images is significantly more complex than that involved in normal images. Therefore, directly utilizing current unpaired optimization algorithms (e.g. generative adversarial learning and circular consistency learning) in underwater sonar image denoising tasks is not sufficient to explore the essential connection from noisy inputs to clean outputs. In this paper, we propose a novel framework for unpaired sonar image denoising which probes the same features of unpaired samples through a dual contrast optimization approach in latent space, named as NCD-GAN. The designed algorithm is composed of two basic parts: Simultaneous Transition Structure (STS) and Separable Contrast Unit (SCU). Separately, STS adopts a cyclic structure of multi-domain transformation to generate valuable surface information, and mines potential feature distributions between different domains via the addition of bidirectional mappings. Meanwhile, SCU introduces additional constraints in layer domains to encourage closer background layers between adjacent domains, while increasing the distance between noise layers so as to better promote the removal of noise and assist in image restoration. Sufficient experiments confirm that our algorithm provides better performance than existing unpaired denoising algorithms on multiple publicly available sonar image datasets and exhibits excellent robustness in the processing of multiple noises. Compared to the latest denoising algorithms, there is at most a 26.43% improvement in the evaluation metrics. Furthermore, it gains a 22.75% mAP boost on the subsequent advanced visual tasks.

Effect of Biogas Slurry Return to Field on Heavy Metal Accumulation in Soil-crop System: A Meta-analysis

To investigate the effects of biogas slurry return-to-field methods, the duration of biogas slurry return to field and the amount of heavy metals brought in from biogas slurry on the accumulation of heavy metals in soil-crop systems, and the importance of factors influencing heavy metal accumulation, 41 papers and 1972 pairs of data were integrated and analyzed. The results showed that the application of biogas slurry alone significantly increased the accumulation of As, Cd, Cr, Cu, and Zn in soil and As and Cr in crops by 20.5%, 15.2%, 25.6%, 18.7%, and 26.3% and 14.6% and 39.5%, respectively, and it had no significant effect on the accumulation of other heavy metals in crops. The combined application of biogas slurry and chemical fertilizers significantly increased the accumulation of soil Cr and Zn by 8.05% and 4.70% and decreased the accumulation of As by crops. Correlation analysis showed that the accumulation rates of soil As, Cd, and Cr were highly significantly and positively correlated (P<0.01) with the duration of biogas slurry return to field and soil organic matter (SOM) content, with correlation coefficients of 0.30, 0.15, and 0.13 and 0.22, 0.27, and 0.22, respectively; they were highly significantly and negatively correlated (P<0.01) with soil pH, with correlation coefficients of 0.16, 0.13, and 0.11, respectively. The heavy metals brought in by biogas slurry return to field promoted the accumulation of As, Cd, and Cr in soil and As, Cd, Cr, and Zn in crops, whereas the accumulation of Cd, Cu, and Zn in soil promoted the accumulation of Cd, Cu, and Zn in crops, with correlation coefficients of 0.45, 0.58, and 0.42, respectively. The main factors of heavy metal accumulation in the soil-crop systems were the duration of biogas slurry return to field, SOM, and soil pH.

A self-validation Noise2Noise training framework for image denoising

ABSTRACT Image denoising is a crucial algorithm in image processing that aims to enhance image quality. Deep learning-based image denoising methods can be categorized into supervised and unsupervised approaches. Supervised learning requires pairs of noisy and noise-free training data, which is impractical in real-world scenarios. Unsupervised learning uses pairs of noisy images for training, but it may yield lower accuracy. Additionally, deep learning-based methods often require a large amount of training data. To overcome these challenges, this research proposes a self-validation Noise2Noise (SV-N2N) framework that generates validation sets using only noisy images without requiring noise-free pairs. The proposed SV-N2N method effectively reduces noise, comparable to supervised and unsupervised methods, without requiring a noise-free ground truth, which is efficient for solving real-world scenarios.

An assessment of dip-slip versus strike-slip faulting modes along the Patagonian Andes (39°-50° S) and their related orogenic models

ABSTRACT The compilation of a database of 225 strain/stress tensors (accounting for ~ 4000 fault-striae data pairs) combined with evidence of syntectonic strata, and a seismotectonic appraisal along the Patagonian Andes (39°-50° S; South America), leads us to assess typical features of Mesozoic to current faulting in this segment of the Andean orogen. At the orogen scale, the fault-slip and focal mechanism database show the prevalence of strike-slip faulting, challenging the classic proposal for the Andean orogen of alternating extensional and contractional deformation phases. However, the wrenching tectonism in Cenozoic times may be explained through a strain partitioning model in the North Patagonian Andes by re-activating a large-scale, inherited anisotropy within the basement (namely the Liquiñe-Ofqui fault system), which is currently the locus of significant seismicity. On the other hand, a non-partitioned context allows explaining the Cenozoic patterns of brittle deformation the South Patagonian Andes. Our results highlight that the classifications of Andean-type orogens should integrate both inherited anisotropies and the heterogeneous distribution of strain across the upper plate in order to address complex patterns of deformation.

Relationship between sonorheometry parameters and laboratory values in a critical care setting in Italy: a retrospective cohort study.

This preliminary retrospective cohort study analyzed the relationship between the parameters provided by sonorheometry device Quantra and the coagulation values obtained from standard venous blood samples in patients admitted in intensive care unit (ICU). We reviewed medical charts of 13 ICU adult patients in whom at least one coagulation study with Quantra was performed. The relationship between Quantra and laboratory data was analyzed with the Spearman rank correlation coefficient (rho). The 95% confidence interval (CI) was computed. A P-value <0.05 was considered statistically significant. We collected 28 data pairs. Statistically significant moderate correlations were found for the following parameters: clot time (CT) and activated partial thromboplastin time (rho=0.516; 95% CI, 0.123-0.904; P=0.009; clot stiffness (CS) and the international normalized ratio (INR; rho=0.418; 95% CI, 0.042-0.787; P=0.039); INR and platelet contribution to CS (rho=0.459; 95% CI, 0.077-0.836; P=0.022); platelet count and platelet contribution to CS (PCS; rho=0.498; 95% CI, 0.166-0.825; P=0.008); and fibrinogen and fibrinogen contribution to CS (FCS; rho=0.620; 95% CI, 0.081-0.881; P=0.001). Quantra can provide useful information regarding coagulation status, showing modest correlations with the parameters obtained from laboratory tests. During diffuse bleeding, CT and FCS values can guide the proper administration of clotting factors and fibrinogens. However, the correlation of INR with CS and PCS can cause misinterpretation. Further studies are needed to clarify the relationship between Quantra parameters and laboratory tests in the critical care setting and the role of sonorheometry in guiding targeted therapies and improving outcomes.

Causal calibration: iteratively calibrating LiDAR and camera by considering causality and geometry

AbstractThe external calibration between 3D LiDAR and 2D camera is an extremely important step towards multimodal fusion for robot perception. However, its accuracy is still unsatisfactory. To improve the accuracy of calibration, we first analyze the interference factors that affect the performance of the calibration model under a causal inference framework in this study. Guided by the causality analysis, we present Iter-CalibNet (Iterative Calibration Convolutional Neural Network) to infer a 6 degrees of freedom (DoF) rigid body transformation between 3D LiDAR and 2D camera. By downscaling point clouds to obtain more overlapping region between 3D–2D data pair and applying iterative calibration manner, the interference of confounding bias in the calibration model is effectively eliminated. Moreover, our Iter-CalibNet adds non-local neural network after each convolution operation to capture the transformation relationship. We also combine the geometric loss and photometric loss obtained from the interframe constraints to optimize the calibration accuracy. Extensive experiments demonstrate that our Iter-CalibNet can achieve leading performance by comparison with other CNN based and traditional calibration methods.

Unsupervised Cross-Domain Rumor Detection with Contrastive Learning and Cross-Attention

Massive rumors usually appear along with breaking news or trending topics, seriously hindering the truth. Existing rumor detection methods are mostly focused on the same domain, thus have poor performance in cross-domain scenarios due to domain shift. In this work, we propose an end-to-end instance-wise and prototype-wise contrastive learning model with cross-attention mechanism for cross-domain rumor detection. The model not only performs cross-domain feature alignment, but also enforces target samples to align with the corresponding prototypes of a given source domain. Since target labels in a target domain are unavailable, we use a clustering-based approach with carefully initialized centers by a batch of source domain samples to produce pseudo labels. Moreover, we use a cross-attention mechanism on a pair of source data and target data with the same labels to learn domain-invariant representations. Because the samples in a domain pair tend to express similar semantic patterns especially on the people’s attitudes (e.g., supporting or denying) towards the same category of rumors, the discrepancy between a pair of source domain and target domain will be decreased. We conduct experiments on four groups of cross-domain datasets and show that our proposed model achieves state-of-the-art performance.

118-LB: Feasibility of a Prototype Dual Function Glucose Sensing Insulin Delivery Cannula in People with Type 1 Diabetes

Background: SynerGTM (Pacific Diabetes Technology, Portland, OR), a prototype subcutaneous cannula combining glucose sensing and insulin infusion could reduce the burden on people with diabetes (PWD) using insulin pump therapy (IPT) and glucose sensors. Aim: To assess and refine SynerGTM sensor accuracy, evaluate insulin infusion site function, and user tolerability. Method: Adults with type 1 diabetes established on IPT participated. Following a 48-hour run-in using a commercially available infusion set with a study pump (Medtronic 780G), a SynerGTM set was inserted subcutaneously in the anterior abdomen and attached to the pump. Participants were blinded to the investigational device readings. A standardised meal was eaten 60 minutes post-insertion. Glucose was measured on arterialised blood by YSI and finger-stick test strips (Roche Accu-Chek Guide) every 10 minutes for 60 minutes before and every 15 minutes for 4 hours after the meal. An insulin bolus was delivered immediately prior to the study meal using SynerGTM. Insulin continued to be delivered using SynerGTM at home and fingerstick readings continued. A second meal test as described above was performed on day 4 before study conclusion. A commercially available (Dexcom G6) continuous glucose monitor (CGM) was inserted prior to the 48 hour run-in preceding day 1 and worn throughout the study. Result: In 15 participants 99% of 679 data pairs of SynerGTM vs. YSI and 1305 data pairs vs. capillary glucose measurement were in zones A and B of the consensus error grid. MARD was 13.5% and 13.5% respectively. No artifact sensor spikes were observed with bolused insulin. SynerGTM was well tolerated and there were no significant differences in CGM time in range or TDD of insulin compared with run-in. Conclusion: This study proves the feasibility of SynerGTM as the first single through-the-skin device providing co-located insulin infusion and glucose sensing for PWD who take insulin. Disclosure J. Lu: None. C. Yuan: None. Y. W. Kong: None. E. Netzer: None. H. Wu: None. T. Seidl: None. D. N. O'neal: None. Funding JDRF (IND-2016-261)

940-P: Real-World Concordance between Point-of-Care and Continuous Glucose Monitoring among Hospitalized Patients with Diabetes

Concordance between continuous glucose monitoring (CGM) and point-of-care testing (POCT) in research studies has been promising; however, the accuracy of using CGM for glucose management in real-world settings requires evaluation. Amid the COVID-19 pandemic, the CGM as “standard of care” (CGM as SOC) program was implemented at Scripps Mercy Hospital in San Diego, CA. In this real-world setting, concordance between the Dexcom G6 CGM system and POCT was evaluated among hospitalized adult patients with diabetes requiring insulin from May 2020-July 2022. For N=880 admissions, CGM readings over the first 24 hours of wear were analyzed relative to POCT conducted within 5 minutes. Patients had a median of 3 CGM-POCT data pairs, and the first CGM-POCT pair occurred a median of 2.75 hours after CGM device placement. Mean absolute relative difference (MARD) decreased over the first 24 hours of wear (B= -0.09%, p=0.0005; Figure 1A). Following the %20/20 rule, 62% of CGM readings were concordant upon the first measured POCT; 15.8% did not meet criteria at all during the first 24-hours. Using the Clarke Error Grid to describe concordance between CGM-POCT pairs (n=3015), only 0.7% (n=22) of CGM readings would have led to incorrect treatment decisions indicated by zones D/E (Figure 1B). These findings provide support for the accuracy of CGM for glucose management in the hospital. Disclosure S.R.Spierling bagsic: None. L.Talavera: None. A.Philis-tsimikas: Advisory Panel; Dexcom, Inc., Novo Nordisk A/S, Sanofi, Other Relationship; Medtronic, Research Support; Novo Nordisk A/S, Lilly, Viking Therapeutics, NIH - National Institutes of Health. A.L.Fortmann: Employee; Dexcom, Inc. R.Belasco: None. A.Ritko: None. S.Lohnes: Consultant; Dexcom, Inc. H.Sandoval: None. M.Chichmarenko: None. E.Soriano: None. A.Bastian: None.

Exhaustive matching of 3D/2D coronary artery structure based on imperfect segmentations.

The 3D/2D coronary artery registration technique has been developed for the guidance of the percutaneous coronary intervention. It introduces the absent 3D structural information by fusing the pre-operative computed tomography angiography (CTA) volume with the intra-operative X-ray coronary angiography (XCA) image. To conduct the registration, an accurate matching of the coronary artery structures extracted from the two imaging modalities is an essential step. In this study, we propose an exhaustive matching algorithm to solve this problem. First, by recognizing the fake bifurcations in the XCA image caused by projection and concatenating the fractured centerline fragments, the original XCA topological structure is restored. Then, the vessel segments in the two imaging modalities are removed orderly, which generates all the potential structures to simulate the imperfect segmentation results. Finally, the CTA and XCA structures are compared pairwise, and the matching result is obtained by searching for the structure pair with the minimum similarity score. The experiments were conducted based on a clinical dataset collected from 46 patients and comprising of 240 CTA/XCA data pairs. And the results show that the proposed method is very effective, which achieves an accuracy of 0.960 for recognizing the fake bifurcations in the XCA image and an accuracy of 0.896 for matching the CTA/XCA vascular structures. The proposed exhaustive structure matching algorithm is simple and straightforward without any impractical assumption or time-consuming computations. With this method, the influence of the imperfect segmentations is eliminated and the accurate matching could be achieved efficiently. This lays a good foundation for the subsequent 3D/2D coronary artery registration task.

Multi-modal sentiment analysis based on graph neural network

Thanks to popularity of social media, people are witnessing the rapid proliferation of posts with various modalities. It is worth noting that these multi-modal expressions share certain characteristics, including the interdependence of objects in the posted images, which is sometimes overlooked in previous researches as they focused on single image-text posts and pay little attention on obtaining the global features. In this paper, a neural network with multiple channels for image-text sentiment detection is proposed. The first step is to encode text and images to capture implicit tendencies. Then the introduction of this model obtains multi-modal expressions by collecting the shared characteristics of the dataset. Finally, the attention mechanism provides reliable predictions of the sentiment tendencies of the given pairs of image-text data. The results of experiments conducted on two publicly available datasets crawled from Twitter prove the reliability of the model on multi-modal sentiment detection, since the model precedes previously proposed models in the main evaluating criteria.

Semiparametric Trend Analysis for Stratified Recurrent Gap Times Under Weak Comparability Constraint

Recurrent event data are frequently encountered in many longitudinal studies where each individual may experience more than one event. Wang and Chen (Biometrics 56(3):789–794, 2000) proposed a comparability constraint to estimate the time trend for the gap times, where the gap time pairs that satisfy the constraint have the same conditional distribution. However, the comparable paired gap times are also independent. Therefore, the comparable gap time pairs will be subject to a stronger constraint than needed for the estimation. Thus their procedure is subject to information loss. Under the accelerated failure time model, we propose a new comparability constraint that can overcome the drawback mentioned above. The gap time pairs being selected by the proposed comparability constraint will still have the same distribution, but they do not need to be independent of each other. We showed that the proposed comparability constraint will utilize more gap time data pairs than the strong comparability. And we showed via various simulation studies that the variance will be smaller than Wang and Chen’s (2000) estimator. We apply the proposed method to the HIV Prevention Trial Network 052 study.

Research on a Rolling Bearing Fault Diagnosis Method Based on Multi-Source Deep Sub-Domain Adaptation

Rolling bearings are the core component of rotating machinery. In order to solve the problem that the distribution of collected rolling bearing data is inconsistent during the operation of bearings under complex working conditions, which results in poor fault identification effects, a fault diagnosis method based on multi-source deep sub-domain adaptation (MSDSA) is proposed in this paper. The proposed method uses CMOR wavelet transform to transform the collected vibration signal into time–frequency maps and construct multiple sets of source–target domain data pairs, and a rolling bearing fault diagnosis network based on a multi-source deep sub-domain adaptive network is established. The network uses shared and domain-specific feature extraction networks to extract data features together. At the same time, the local maximum mean discrepancy (LMMD) was introduced to effectively capture the fine-grained information of the category. Each set of data was used to train the corresponding classifier. Finally, multiple sets of classifiers were combined to reduce the classification loss of the target domain samples at the classification boundary to achieve fault identification. In order to make the training process more stable, the network used the Ranger optimizer for parameter tuning. This paper verifies the effectiveness of the proposed method through two sets of comparative experiments. The proposed method achieves 97.78%, 99.65%, and 99.34% in three migration tasks. The experimental results show that the proposed method has a high recognition rate and generalization performance.

Adaptive Maximum Entropy Graph-Guided Fast Locality Discriminant Analysis.

Linear discriminant analysis (LDA) aims to find a low-dimensional space in which data points in the same class are to be close to each other while keeping data points from different classes apart. To improve the robustness of LDA to non-Gaussian distribution data, most existing discriminant analysis methods extend LDA by approximating the underlying manifold of data. However, these methods suffer from the following problems: 1) local affinity or reconstruction coefficients are learned on the basis of the relationships of all data pairs, which would lead to a sharp increase in the amount of computation and 2) they learn the manifold information in the original space, ignoring the interference of the noise and redundant features. Motivated by these challenges, this article represents a novel discriminant analysis model, called fast and adaptive locality discriminant analysis (FALDA), to improve the efficiency and robustness. First, with the anchor-based strategy, a bipartite graph of each class is constructed to characterize the local structure of data. Since the number of anchor points is far less than that of data points, learning of fuzzy membership relationships between data points and anchor points within each class can save training time. Second, a maximum entropy regularization is introduced to control the uniformity of the weights of graphs and avoid the trivial solution. Third, the above relationships are updated adaptively in the process of dimensionality reduction, which can suppress the interference of the noise and redundant features. Fourth, the whitening constraint is imposed on the projection matrix to remove the relevance between features and restrict the total scatter of data in the subspace. Last but not the least, data with complex distribution can be explicitly divided into sub-blocks according to the learned anchor points (or subclass center points). We test our proposed method on synthetic data, benchmark datasets, and imbalanced datasets. Promising experimental results demonstrate the success of this novel model.

Nitrogen Application Effect on Maize Yield, NH3, and N2O Emissions in Northeast China by Meta-Analysis

Ammonia volatilization (NH3) and nitrous oxide (N2O) emission are the main underliers of nitrogen loss in farmlands, which can decrease nitrogen use efficiency and trigger environmental problems regarding greenhouse effects. Previous studies have regional limitations and lack universal guiding significance, as they are primarily based on individual experiments related to the effects of applying nitrogen fertilizer on maize. In this study, we performed a meta-analysis on a regional scale to investigate the overall impact of nitrogen fertilizer application on maize yield, N2O, and NH3 in Northeast China. The database included 85 studies and 1147 pairs of experimental data that were analyzed. The results showed that applying nitrogen fertilizer significantly increased maize yield, N2O emissions, and NH3 volatilization effects, and the growth rates (E) were 50.64%, 64.39%, and 69.25%, respectively. In Northeast China, the average emission factors of N2O and NH3 were 0.72% and 8.21%, respectively. The optimum nitrogen application rate for maize in Northeast China was 205 kg ha−1, resulting in 8.37% nitrogen loss (through N2O and NH3). Soil texture, alkaline nitrogen (AN) content in the soil, mean annual precipitation (MAP), nitrogen application rate, and fertilizer type were the key influential factors affecting changes in maize yield and N loss (N2O and NH3). Yield-scaled N2O and NH3 were found to be the significant emission reduction parameters that ensured maize yield. However, there was a remarkable ‘seesaw effect’ between yield-scaled N2O and NH3 under the same natural conditions (MAP and soil texture). Therefore, human activities such as reducing N surplus in soil, and N fertilizer application rate, along with selecting suitable fertilizer types should be given more attention to reduce yield-scaled N2O and NH3. Moreover, minimizing NH3 and N2O dual emission should be the main objective for green agriculture in Northeast China, rather than over-emphasizing on single emission reduction.

Ground Truth Inference for Weakly Supervised Entity Matching

Entity matching (EM) refers to the problem of identifying pairs of data records in one or more relational tables that refer to the same entity in the real world. Supervised machine learning (ML) models currently achieve state-of-the-art matching performance; however, they require a large number of labeled examples, which are often expensive or infeasible to obtain. This has inspired us to approach data labeling for EM using weak supervision. In particular, we use the labeling function abstraction popularized by Snorkel, where each labeling function (LF) is a user-provided program that can generate many noisy match/non-match labels quickly and cheaply. Given a set of user-written LFs, the quality of data labeling depends on a labeling model to accurately infer the ground-truth labels. In this work, we first propose a simple but powerful labeling model for general weak supervision tasks. Then, we tailor the labeling model specifically to the task of entity matching by considering the EM-specific transitivity property. The general form of our labeling model is simple while substantially outperforming the best existing method across ten general weak supervision datasets. To tailor the labeling model for EM, we formulate an approach to ensure that the final predictions of the labeling model satisfy the transitivity property required in EM, utilizing an exact solution where possible and an ML-based approximation in remaining cases. On two single-table and nine two-table real-world EM datasets, we show that our labeling model results in a 9% higher F1 score on average than the best existing method. We also show that a deep learning EM end model (DeepMatcher) trained on labels generated from our weak supervision approach is comparable to an end model trained using tens of thousands of ground-truth labels, demonstrating that our approach can significantly reduce the labeling efforts required in EM.

Constructing a link between multivariate titanium-based semiconductor band gaps and chemical formulae based on machine learning

Titanium-based semiconductors are wildly recognized as one of the most commonly used photocatalysts for photocatalysis. Energy band modulation is a key aspect of the catalytic activity of photocatalytic semiconductors, but the acquisition of semiconductor energy bands is still a complex and important task. In recent years, machine learning has played an important role in materials prediction, where the crystal structure of a material is usually used as input in energy band prediction. However, existing machine learning algorithms cannot accurately predict the energy bands of materials from structural components. Here, we convert the chemical formula into component descriptors after comparing first principles and ultraviolet-visible spectrophotometry (UV–vis) errors on bandgap values, and the component and bandgap values form a set of labeled data pairs. The chemical formula components are used as input to accurately predict the energy bands of the material. In our evaluation, the model outperforms existing machine learning methods in predicting energy bands, yielding mean absolute value errors of about 0.277 eV, and possesses a significant advantage in component prediction. In particular, this method of predicting the photocatalytic semiconductor energy band gap from chemical formula components provides a new way of thinking about photocatalyst selectivity.

Comparison of record linkage methods

Record linkage is an important tool to enhance database integration. This is even more valuable in a scenario with more hefty budget cuts and a growing drop in response rate in traditional surveys. This strategy makes it possible to expand the crossing alternatives with variables not present in the original base. However, there are many different data pairing methods exposed in the literature. In this sense, the objective of this paper is to compare well-known methods of record linkage. The comparison was made in synthetic dataset. To compare the methods, it was adopted a quantitative approach based on the Precision, Recall, and F-Statistics metrics, using two comparison functions: Levenshtein and Jaro-Winkler. Among the six types of classifiers analyzed, the supervised methods had the best results.