关键词： NO2   surface concentration   TROPOMI   satellite   Germany   machine learning   exposure   COVID-19  

摘要： In this paper, we present the estimation of surface NO2 concentrations over Germany using a machine learning approach. TROPOMI satellite observations of tropospheric NO2 vertical column densities (VCDs) and several meteorological parameters are used to train the neural network model for the prediction of surface NO2 concentrations. The neural network model is validated against ground-based in situ air quality monitoring network measurements and regional chemical transport model (CTM) simulations. Neural network estimation of surface NO2 concentrations show good agreement with in situ monitor data with Pearson correlation coefficient (R) of 0.80. The results also show that the machine learning approach is performing better than regional CTM simulations in predicting surface NO2 concentrations. We also performed a sensitivity analysis for each input parameter of the neural network model. The validated neural network model is then used to estimate surface NO2 concentrations over Germany from 2018 to 2020. Estimated surface NO2 concentrations are used to investigate the spatio-temporal characteristics, such as seasonal and weekly variations of NO2 in Germany. The estimated surface NO2 concentrations provide comprehensive information of NO2 spatial distribution which is very useful for exposure estimation. We estimated the annual average NO2 exposure for 2018, 2019 and 2020 is 15.53, 15.24 and 13.27 mu g/m(3), respectively. While the annual average NO2 concentration of 2018, 2019 and 2020 is only 12.79, 12.60 and 11.15 mu g/m(3). In addition, we used the surface NO2 data set to investigate the impacts of the coronavirus disease 2019 (COVID-19) pandemic on ambient NO2 levels in Germany. In general, 10-30% lower surface NO2 concentrations are observed in 2020 compared to 2018 and 2019, indicating the significant impacts of a series of restriction measures to reduce the spread of the virus.

关键词： artificial terrace mapping   object-based image analysis (OBIA)   machine learning (ML)   random forest (RF)  

摘要： Artificial terraces are of great importance for agricultural production and soil and water conservation. Automatic high-accuracy mapping of artificial terraces is the basis of monitoring and related studies. Previous research achieved artificial terrace mapping based on high-resolution digital elevation models (DEMs) or imagery. As a result of the importance of the contextual information for terrace mapping, object-based image analysis (OBIA) combined with machine learning (ML) technologies are widely used. However, the selection of an appropriate classifier is of great importance for the terrace mapping task. In this study, the performance of an integrated framework using OBIA and ML for terrace mapping was tested. A catchment, Zhifanggou, in the Loess Plateau, China, was used as the study area. First, optimized image segmentation was conducted. Then, features from the DEMs and imagery were extracted, and the correlations between the features were analyzed and ranked for classification. Finally, three different commonly-used ML classifiers, namely, extreme gradient boosting (XGBoost), random forest (RF), and k-nearest neighbor (KNN), were used for terrace mapping. The comparison with the ground truth, as delineated by field survey, indicated that random forest performed best, with a 95.60% overall accuracy (followed by 94.16% and 92.33% for XGBoost and KNN, respectively). The influence of class imbalance and feature selection is discussed. This work provides a credible framework for mapping artificial terraces.

关键词： Time-dependent reliability   Life-cycle analysis   Machine learning   Patient rule-induction   Polynomial regression   Support vector machine  

摘要： Life-cycle management under uncertainty relies on the determination of life-cycle failure probability and risk. Usually, both life-cycle failure probability and risk are approximated by adding up the annual quantities during the service life of a structure. This approximation is based on the assumptions of (a) low annual failure probabilities and (b) independence of annual failure events. As a structure continuously deteriorates over time, both assumptions are likely to be violated. Therefore, it is crucial to investigate the error associated with this approximation so that the downstream management decisions such as inspection planning and maintenance optimization can be well grounded. In this paper, the error of approximate life-cycle failure probability and risk is analyzed considering different distribution types of structural capacity and demand as well as various deterioration mechanisms. Several machine learning methods are used for this purpose. Specifically, conditions under which the error is acceptable are identified in qualitative analysis using patient rule-induction method (PRIM). Quantitative error analysis based on polynomial and support vector regression is conducted to develop error correction techniques suitable for life-cycle analysis and management. The results of the error analysis are applied in the life-cycle analysis of a deteriorating structure to demonstrate the magnitude of approximation error and the importance of error correction.

关键词： EBA4SUB   Conceptual event-based method   Machine learning   Rainfall-runoff modeling  

摘要： The applicability of four machine learning (ML) methods, ANFIS-PSO, ANFIS-FCM, MARS and M5Tree, together with multi model simple averaging (MM-SA) ensemble method, is investigated in rainfall-runoff modeling at hourly timescale. The results are compared with the conceptual EBA4SUB model using rainfall and runoff data from Samoggia River basin, Italy. The capability of the methods is measured using five statistics, Nash-Sutcliffe efficiency, root mean squared error, mean absolute error, scatter index, and adjusted index of agreement. Comparison of single ML reveals that the ANFIS-PSO, ANFIS-FCM and MARS produce similar accuracy which is better than the M5Tree model. MM-SA ensemble model improves the accuracy of ANFIS-PSO, ANFIS-FCM, MARS and M5Tree models with respect to RMSE by 8.5%, 5%, 7.4% and 28.8%, respectively. Comparison with the conceptual event-based method indicates that the ML methods generally performs superior to the EBA4SUB;however, latter method provides better accuracy than the M5Tree and MARS in some cases. [GRAPHICS] .

关键词： Turbulence  

摘要： This work presents a review of the current state of research in data-driven turbulence closure modeling. It offers a perspective on the challenges and open issues but also on the advantages and promises of machine learning (ML) methods applied to parameter estimation, model identification, closure term reconstruction, and beyond, mostly from the perspective of large Eddy simulation and related techniques. We stress that consistency of the training data, the model, the underlying physics, and the discretization is a key issue that needs to be considered for a successful ML-augmented modeling strategy. In order to make the discussion useful for non-experts in either field, we introduce both the modeling problem in turbulence as well as the prominent ML paradigms and methods in a concise and self-consistent manner. In this study, we present a survey of the current data-driven model concepts and methods, highlight important developments, and put them into the context of the discussed challenges. © 2021 The Authors. GAMM - Mitteilungen published by Wiley-VCH GmbH.

关键词： Minimum ignition energy (MIE)   Quantitative structure-property relationship   (QSPR)   Machine learning  

摘要： Quantitative Structure-Property Relationship (QSPR) regression models for Minimum Ignition Energy (MIE) prediction of 60 flammable compounds have been developed using machine learning algorithms Random Forests (RF), Artificial Neural Networks (ANN) and a non-machine learning algorithm Genetic Function Approximation (GFA). RF algorithm was implemented for feature selection to identify the 13 molecular descriptors having maximum effect on MIE prediction accuracy (i.e. descriptors affecting MIE prediction accuracy > 1%). Thereafter, these descriptors were used as input parameters to develop the RF, ANN and GFA models. The optimized RF algorithm resulted in test set R2 of 0.85 and displayed high internal robustness and external predictability. The ANN and GFA algorithms displayed improved performance only on addition of an additional parameter Structure Parameter to the existing 13 descriptor set. The optimized ANN and the GFA models displayed a test set R2 of 0.79 and 0.71, respectively. The ANN model resulted in lower machine learning bias as compared to the RF model. The ANN model was observed to be more reliable than RF model for MIE prediction of small datasets. Based on this work, the RF algorithm for feature selection in QSPR modeling and the ANN, RF algorithms were observed to be promising options for MIE prediction applications.

关键词： coeliac disease   gluten   T‐   lymphocyte   T‐   cell receptor repertoire   machine learning   TRG   TRD   clustering   duodenum  

摘要： In coeliac disease (CeD), immune-mediated small intestinal damage is precipitated by gluten, leading to variable symptoms and complications, occasionally including aggressive T-cell lymphoma. Diagnosis, based primarily on histopathological examination of duodenal biopsies, is confounded by poor concordance between pathologists and minimal histological abnormality if insufficient gluten is consumed. CeD pathogenesis involves both CD4(+) T-cell-mediated gluten recognition and CD8(+) and gamma delta T-cell-mediated inflammation, with a previous study demonstrating a permanent change in gamma delta T-cell populations in CeD. We leveraged this understanding and explored the diagnostic utility of bulk T-cell receptor (TCR) sequencing in assessing duodenal biopsies in CeD. Genomic DNA extracted from duodenal biopsies underwent sequencing for TCR-delta (TRD) (CeD, n = 11;non-CeD, n = 11) and TCR-gamma (TRG) (CeD, n = 33;non-CeD, n = 21). We developed a novel machine learning-based analysis of the TCR repertoire, clustering samples by diagnosis. Leave-one-out cross-validation (LOOCV) was performed to validate the classification algorithm. Using TRD repertoire, 100% (22/22) of duodenal biopsies were correctly classified, with a LOOCV accuracy of 91%. Using TCR-gamma (TRG) repertoire, 94.4% (51/54) of duodenal biopsies were correctly classified, with LOOCV of 87%. Duodenal biopsy TRG repertoire analysis permitted accurate classification of biopsies from patients with CeD following a strict gluten-free diet for at least 6 months, who would be misclassified by current tests. This result reflects permanent changes to the duodenal gamma delta TCR repertoire in CeD, even in the absence of gluten consumption. Our method could complement or replace histopathological diagnosis in CeD and might have particular clinical utility in the diagnostic testing of patients unable to tolerate dietary gluten, and for assessing duodenal biopsies with equivocal features. This approach is generalisa

关键词： Carbon nanotubes   nanocomposites   machine learning   tensile strength   Gaussian process regression  

摘要： This paper is devoted to the development and construction of a practical Machine Learning (ML)-based model for the prediction of tensile strength of polymer carbon nanotube (CNTs) composites. To this end, a database was compiled from the available literature, composed of 11 input variables. The input variables for predicting tensile strength of nanocomposites were selected for the following main reasons: (i) type of polymer matrix, (ii) mechanical properties of polymer matrix, (iii) physical characteristics of CNTs, (iv) mechanical properties of CNTs and (v) incorporation parameters such as CNT weight fraction, CNT surface modification method and processing method. As the problem of prediction is highly dimensional (with 11 dimensions), the Gaussian Process Regression (GPR) model was selected and optimized by means of a parametric study. The correlation coefficient (R), Willmott's index of agreement (IA), slope of regression, Mean Absolute Percentage Error (MAPE), Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) were employed as error measurement criteria when training the GPR model. The GPR model exhibited good performance for both training and testing parts (RMSE = 5.982 and 5.327 MPa, MAE = 3.447 and 3.539 MPa, respectively). In addition, uncertainty analysis was also applied to estimate the prediction confidence intervals. Finally, the prediction capability of the GPR model with different ranges of values of input variables was investigated and discussed. For practical application, a Graphical User Interface (GUI) was developed in Matlab for predicting the tensile strength of nanocomposites.

关键词： PAH   Wastewater treatment sludge   UV-C light   Data mining   machine learning   Over-sampling methods   Prediction of PAH removal efficiency  

摘要： Removal of polycyclic aromatic hydrocarbons (PAHs) from wastewater treatment sludge with appropriate technologies is of great importance for nature and public health. UV technology is one of the most frequently used methods for the removal of PAHs. While various photodegradation applications with UV-C (ultraviolet-C) light and photocatalysts can be performed to remove these compounds, a large number of tests should be implemented to determine optimum removal conditions, which increase time and cost. It is possible to make predictions for the removal efficiency of PAHs by using data mining classification and reveal the hidden knowledge from data. This study aims to determine appropriate machine learning (ML) methods for the prediction of the PAH removal efficiency from wastewater treatment sludges regarding the initial PAH levels. The samples have multi-class imbalanced outputs;thus, random over-sampling and Synthetic Minority Over-sampling TEchniques (SMOTE) are used to improve the prediction results. Well-known data mining classification/machine learning methods, artificial neural network (multi-layer perceptron-MLP), k-means (k-NN), support vector machine (SVM), decision tree (C4.5), random forest (RF), and Bagging, are proposed for the prediction of removal efficiencies. Different evaluation metrics, Accuracy, multi-class AUC (MAUC-multi-class area under ROC curve), F-measure, Precision, Recall, and Specificity are used for the performance comparisons. RF and k-NN perform better with 92.35% and 92.36% average prediction accuracies, respectively. Besides, RF outperforms other methods with 0.97 MAUC value. RF and k-NN can be used for the removal efficiency prediction on the multi-class imbalanced datasets successfully, and removal efficiencies can be highly predicted considering input components with less cost and effort.

关键词： Machine learning   Decision making   Decision tree   Random forest   Automated neural network  

摘要： Background Machine learning is a broad term encompassing a number of methods that allow the investigator to learn from the data. These methods may permit large real-world databases to be more rapidly translated to applications to inform patient-provider decision making. Methods This systematic literature review was conducted to identify published observational research of employed machine learning to inform decision making at the patient-provider level. The search strategy was implemented and studies meeting eligibility criteria were evaluated by two independent reviewers. Relevant data related to study design, statistical methods and strengths and limitations were identified;study quality was assessed using a modified version of the Luo checklist. Results A total of 34 publications from January 2014 to September 2020 were identified and evaluated for this review. There were diverse methods, statistical packages and approaches used across identified studies. The most common methods included decision tree and random forest approaches. Most studies applied internal validation but only two conducted external validation. Most studies utilized one algorithm, and only eight studies applied multiple machine learning algorithms to the data. Seven items on the Luo checklist failed to be met by more than 50% of published studies. Conclusions A wide variety of approaches, algorithms, statistical software, and validation strategies were employed in the application of machine learning methods to inform patient-provider decision making. There is a need to ensure that multiple machine learning approaches are used, the model selection strategy is clearly defined, and both internal and external validation are necessary to be sure that decisions for patient care are being made with the highest quality evidence. Future work should routinely employ ensemble methods incorporating multiple machine learning algorithms.