Search

This document is a guest editorial from the journal IET Generation, Transmission & Distribution. It discusses the use of artificial intelligence (AI) in reliable forecasting for energy sectors. The editorial highlights the challenges of integrating renewable energy sources and fluctuating electricity demand, and emphasizes the importance of accurate forecasting for system operators. The document also provides summaries of several papers included in a special issue on AI-empowered forecasting in energy sectors, covering topics such as load forecasting, wind power prediction, and control parameter optimization. The editorial concludes by recommending further research and practical implementations of AI approaches in the energy sectors. [Extracted from the article]

The reversible pump‐turbine plays an important role in hydropower stations, but pressure pulsation during their operation affects their performance and lifespan. Accurate prediction of pressure pulsation signals can provide an important basis for energy planning and stable operation of pumped storage units, thereby promoting sustainable development of the environment. This study introduces an optimization method that combines long short‐term memory (LSTM) and variable mode decomposition (VMD) to enhance the prediction accuracy of pressure pulsation signals. First, by decomposing the pressure pulsation signal into multiple relatively stable subsequence components using VMD, the characteristics of the original signal become more distinct. Subsequently, individual LSTM‐based time series prediction models were constructed for each modal function, and the hyperparameters related to subsequence were optimized using the sparrow search algorithm. To validate the efficacy of the proposed approach, this paper conducted experiments using pressure pulsation signals of a pump‐turbine obtained through numerical simulation. The experimental data was divided into training and testing sets, with the former used to train the LSTM model and the latter used for validation. The experimental results show that the optimized VMD with an optimized LSTM method can effectively improve the prediction accuracy of pressure pulsation signals in reversible pump‐turbine. [ABSTRACT FROM AUTHOR]

Long‐term creep compliance is one of the most important mechanical properties for evaluating the long‐term behavior of concrete structures. This paper aims to optimize machine‐learning models to predict this viscoelastic property. The most relevant dataset available in the literature is considered, cleaned, and preprocessed to optimize the outcome. The advanced XGBoost model, which is to be the most effective shallow machine‐learning model for modeling tabular datasets, is employed in this study to maximize model accuracy. Short‐term creep compliances of a given sample at typical ages are used as input features to model the long‐term creep compliance of concrete. This approach outperforms standard machine‐learning approaches that do not include short‐term creep as an input feature. Indeed, the short‐term behavior of concrete strongly influents its long‐term one. The optimized machine model presented herein is accurate and useful for practical applications. It uses input features that are easy to obtain to predict long‐term creep compliance up to several decades, which is difficult and expensive to measure. [ABSTRACT FROM AUTHOR]

LINKED CONTENT: This article is linked to Verma et al papers. To view these articles, visit https://doi.org/10.1111/apt.17891 and https://doi.org/10.1111/apt.17913 [ABSTRACT FROM AUTHOR]

Software maintenance is an important phase in the software development life cycle. Software projects maintain bug repositories to gather, organize, and keep track of bug reports. These bug reports are resolved by numerous software developers. Whenever the reported bug does not get resolved by the assigned developer, he marks the resolution of bug report as Non‐Reproducible (NR). When NR bugs are reconsidered, few of them get resolved, and their resolution changes from NR to fix (NRF). The main aim of this paper is to predict these fixable NRF bug reports. A major challenge in predicting NRF bugs from NR bugs is that only a small portion of NR bugs get fixed, i.e., class‐imbalance problem. For example, NRF bugs account for only 8.64%, 4.73 %, 4.56%, and 1.06% in NetBeans, Eclipse, Open Office, and Mozilla Firefox projects respectively. In this paper, we work on improving the classification performance on these imbalanced datasets. We propose IMNRFixer, a novel and hybrid NRF prediction tool. IMNRFixer uses three different techniques to combat class‐imbalance problem: undersampling, oversampling, and ensemble models. We evaluate the performance of IMNRFixer models on four large and open‐source projects of Bugzilla repository. Our results show that IMNRFixer outperforms conventional machine learning techniques. IMNRFixer achieves performance up to 71.7%, 93.1%, 91.7%, and 96.5% while predicting the minority class (NRF) for NetBeans, Eclipse, Open Office, and Mozilla Firefox projects, respectively. [ABSTRACT FROM AUTHOR]

Remaining driving range (RDR) research has continued to consistently evolve with the development of electric vehicles (EVs). Accurate RDR prediction is a promising approach to alleviate distance anxiety when power battery technology is not yet fully matured. This paper first introduces the research motivation of RDR prediction, summarizes the previous research progress, and classifies the influencing factors of RDR. Second, conduct research and analysis on the physical model of EVs, mainly including battery and vehicle models. Based on the physical model, the energy flow problem of EVs is analyzed and discussed. Third, four key challenges of RDR prediction are summarized: battery state estimation, driving behavior classification and recognition, driving condition prediction and speed prediction, and RDR calculation method. Finally, given the four challenges faced by RDR, a driving range prediction method based on vehicle‐cloud collaboration is proposed, which combines the advantages of cloud computing and machine learning to provide further research trends. [ABSTRACT FROM AUTHOR]

Spatial econometric models estimated on the big geo‐located point data have at least two problems: limited computational capabilities and inefficient forecasting for the new out‐of‐sample geo‐points. This is because of spatial weights matrix W defined for in‐sample observations only and the computational complexity. Machine learning models suffer the same when using kriging for predictions; thus this problem still remains unsolved. The paper presents a novel methodology for estimating spatial models on big data and predicting in new locations. The approach uses bootstrap and tessellation to calibrate both model and space. The best bootstrapped model is selected with the PAM (Partitioning Around Medoids) algorithm by classifying the regression coefficients jointly in a nonindependent manner. Voronoi polygons for the geo‐points used in the best model allow for a representative space division. New out‐of‐sample points are assigned to tessellation tiles and linked to the spatial weights matrix as a replacement for an original point what makes feasible usage of calibrated spatial models as a forecasting tool for new locations. There is no trade‐off between forecast quality and computational efficiency in this approach. An empirical example illustrates a model for business locations and firms' profitability. [ABSTRACT FROM AUTHOR]

Computational methods have gained prominence in healthcare research. The accessibility of healthcare data has greatly incited academicians and researchers to develop executions that help in prognosis of cancer drug response. Among various computational methods, machine‐learning (ML) and deep‐learning (DL) methods provide the most consistent and effectual approaches to handle the serious aftermaths of the deadly disease and drug administered to the patients. Hence, this systematic literature review has reviewed researches that have investigated drug discovery and prognosis of anticancer drug response using ML and DL algorithms. Fot this purpose, PRISMA guidelines have been followed to choose research papers from Google Scholar, PubMed, and Sciencedirect websites. A total count of 105 papers that align with the context of this review were chosen. Further, the review also presents accuracy of the existing ML and DL methods in the prediction of anticancer drug response. It has been found from the review that, amidst the availability of various studies, there are certain challenges associated with each method. Thus, future researchers can consider these limitations and challenges to develop a prominent anticancer drug response prediction method, and it would be greatly beneficial to the medical professionals in administering non‐invasive treatment to the patients. [ABSTRACT FROM AUTHOR]

Commercial buildings are consuming an increasing amount of energy, and accurate load demand forecasting is critical for the reliable operation of power systems and the efficient use of resources. Therefore, in this paper, a short‐term commercial load forecasting model based on tunicate swarm algorithm (TSA) combined with an extreme learning machine (ELM) under peak‐valley features is proposed as a research case for a shopping mall in Romania. This paper's overall structure is divided into two steps. In the first step, the 24‐h day is divided into six periods by analyzing the daily load characteristics of the training set, and the peak and valley loads are obtained. The ELM optimized by TSA (TSA‐ELM) algorithm is then used to forecast the peak and valley values of the test set one day ahead. In the second step, the actual load, peak, and valley for the previous week of historical load are chosen using the maximum information coefficient (MIC). Following that, the MIC ≥ 0.8 features are added to the TSA‐ELM to achieve short‐term commercial electricity load forecasting. The results show that the PV (Peak & Valley)‐TSA‐ELM model proposed in this paper has higher prediction accuracy compared with other models. Taking ELM as an example, compared with the traditional ELM, the mean absolute error, root mean square error, and mean absolute percentage error of PV‐TSA‐ELM are reduced by 20.59%, 20.13%, and 19.19% on average in the three commercial data sets. The proposed model is validated with an industrial data set, and ideal results are obtained, which verifies the effectiveness and superiority of the proposed method. [ABSTRACT FROM AUTHOR]

Dynamic IR drop prediction based on machine learning has been studied in recent years. However, most proposed models used all input features extracted from circuits or manually selected parts of raw features as inputs, which failed to differentiate the order of priority among input features in a flexible manner. In this paper, QuantumForest to vector‐based dynamic IR drop prediction is introduced. With the sparse attention mechanism brought by QuantumForest, important attributes of circuits are weighed more heavily than others. A new multi‐view feature creation method is also proposed and a novel regional distance feature is built up subsequently. The performance is evaluated on two chip designs with real simulation vectors. The experiment results indicate that the prediction result of the method outperforms other prominent methods for dealing with machine learning based IR drop analysis, reaching an average MAE of only 1.457 mV$\text{mV}$ on two designs. [ABSTRACT FROM AUTHOR]

Polymer gears have long been used on power transmissions with the fundamental durability data, including fatigue S‐N curves, yielding important data informing reliable and compact designs. This paper proposed a prediction method for polyformaldehyde (POM) gear fatigue life based on the innovative WGAN‐XGBoost algorithm. The findings generated herein revealed that the proposed method performs well in terms of prediction accuracy. The predicted fatigue lives, analyzed under different loading conditions, were within 1.5 times dispersion band compared with experimental results. Furthermore, based upon the enhanced fatigue dataset, a thermo‐mechanical coupled prediction formula for POM gear contact fatigue life was proposed. These findings offered a new approach for high‐power density design of polymer gears. Highlights: A prediction method for POM gear fatigue life based on the WGAN‐XGBoost is proposed.A thermo‐mechanical coupled fatigue life prediction formula for POM gears is proposed.The method could shorten the sample size by 25% with the error band within 1.5. [ABSTRACT FROM AUTHOR]

Summary: The rapid update of computing power leads to exponential data traffic growth, and the incidence of network attacks is also increasing. It is significantly important to analyze and predict network traffic accurately in the early stage and take corresponding preventive measures. The existing network flow integrated forecasting models still have some bottlenecks that are difficult to solve, for example, the slow optimization speed of modal decomposition parameters, easy falling into local optimal solutions, the slow convergence speed of the training process, and poor generalization capability. In this paper, particle swarm optimization (PSO) is utilized to improve the parameters selection process of the variational mode decomposition (VMD) algorithm and the extreme learning machine (ELM) algorithm. First, the PSO‐VMD combined with multi‐scale permutation entropy (MPE) is utilized to decompose the original network flow, and multiple eigenmode components are obtained. Second, the PSO‐ELM is utilized to train the network traffic prediction model, and the PSO parameters in PSO‐ELM are updated through adaptive weight adjustment and synchronous learning factors to increase the training and prediction speed, and the component prediction results are reconstructed to get a high‐precision network flow forecasting result. Finally, through the prediction and verification of the public network flow data of the WIDE backbone, the result of this experiment indicates that the VMD‐PSO‐ELM can break through the bottlenecks of slow optimization speed of VMD decomposition parameters, reduce the computational complexity of ELM, accelerate the convergence speed, and increase the forecasting accuracy. [ABSTRACT FROM AUTHOR]

Accurate prediction of remaining useful life (RUL) plays an important role in the formulation of maintenance strategies. However, due to the diversity of the failure mode of equipment, there are significant differences between the degradation data, which greatly affects the accuracy of RUL prediction. In this case, an ensemble prediction model considering health index‐based (HI‐based) classification is proposed in this paper. Firstly, the stacked autoencoder (SAE) is employed to construct the HI. Then, the time window is used to sequentially process the HI sequence, so that many data segments with the same length can be achieved. To differentiate the data with the similar degradation process, K‐means and Xgboost are selected to construct offline and online data classification models respectively. Finally, according to the results of the data classification, the ensemble model that integrates multiple machine learning methods are separately trained and then adaptively used for RUL prediction. In addition, integrating multiple methods helps to improve the generalization ability of the model. The NASA C‐MAPSS dataset is applied to verify the effectiveness of the proposed method, and the results show that the proposed method achieves a higher prediction accuracy and shorter computational time than other existing prediction models. [ABSTRACT FROM AUTHOR]

Nonlinear optical crystals have been a key material in lasers due to their excellent frequency conversion properties. The search for new nonlinear crystals has been extremely active due to the inherent drawbacks of existing nonlinear crystals, such as the low laser damage threshold of infrared nonlinear crystals and the phenomenon of two‐photon absorption. The stability of a material can be characterized by calculating the formation energy, and machine learning (ML) is currently the most mainstream tool to explore materials. In this paper, the digital features of nonlinear crystals are established by using only the component information. Based on the low‐variance and cross‐validated recursive feature elimination (RFECV), 11 features with strong correlation are obtained for model training. Among them, the two features related to electronegativity account for 71.44% of all features, and become the most critical factor for the formation energy predicted by the model. Three ML regression models have been proposed to predict the formation energy of nonlinear crystals. Among them, the gradient boosting regression model shows excellent prediction performance with R2 = 0.939, RMSE = 0.205, and MAE = 0.132 eV per atom. The model provides a useful tool for fast and low‐cost prediction of nonlinear crystal formation energy. [ABSTRACT FROM AUTHOR]

Global carbon dioxide emissions have become a great threat to economic sustainability and human health. The carbon market is recognized as the most promising mean to curb carbon emissions, furthermore, carbon price forecasting will promote the role of the carbon market in emissions reduction and achieve reduction targets at lower economic costs for emission entities. However, there are still some technical problems in carbon price prediction, such as mode mixing and larger reconstruction error for the traditional empirical mode decomposition‐type models. Therefore, the innovation of this paper is constructing a novel carbon price prediction model of complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN)‐long short‐term memory (LSTM), that combines the advantages of CEEMDAN in decomposing the multiscale time‐frequency carbon price signals and the LSTM model in fitting the financial signals. The results show the proposed CEEMDAN‐LSTM model has significant accuracy in predicting the complex carbon price signals. The prediction error and expectation indicators of root mean square error, mean absolute error, mean absolute percentage error, and direction accuracy are 0.638342, 0.448695, 0.015666, and 0.687631, respectively, which is better than other benchmark models. Further evidence convince that the short‐term forecasting performance is superior to the long‐term and medium‐term performance. That evidence concludes that the proposed model is a reliable method to reveal the carbon price‐driving mechanism from the point of multiscale time‐frequency characteristics. Particularly, short‐term forecasting is more accurate and can provide a valuable technical reference for reduction entities and green financial companies to judge the market situation and formulate quantitative transactions. [ABSTRACT FROM AUTHOR]

Compressive strength is considered to be one of the most important mechanical properties of high‐strength concrete (HSC). In this study, three machine learning models, ELM, PSO‐ANN, and GS‐SVR were employed to predict the compressive strength of HSC using 681 data records. The five ingredients and the compressive strength of HSC were regarded as input variable features and output target, respectively. Results indicated that the GS‐SVR model showed the best performance in forecasting with the R of 0.992, MAPE of 0.016, RMSE of 1.241, MAE of 0.842, and RRMSE of 0.024, which could be an alternative candidate for strength prediction. Additionally, a sensitivity analysis was performed within the GS‐SVR model to achieve an in‐depth examination of the effect of each single input variable on the compressive strength. The sensitivity of input variables demonstrated that the water is the most sensitive parameter to the compressive strength of HSC, while the superplasticizer is less sensitive. The research of this paper can provide reference and guidance for concrete design and construction. [ABSTRACT FROM AUTHOR]

Background: Dose calculations for novel radiotherapy cancer treatments such as proton minibeam radiation therapy is often done using full Monte Carlo (MC) simulations. As MC simulations can be very time consuming for this kind of application, deep learning models have been considered to accelerate dose estimation in cancer patients. Purpose: This work systematically evaluates the dose prediction accuracy, speed and generalization performance of three selected state‐of‐the‐art deep learning models for dose prediction applied to the proton minibeam therapy. The strengths and weaknesses of those models are thoroughly investigated, helping other researchers to decide on a viable algorithm for their own application. Methods: The following recently published models are compared: first, a 3D U‐Net model trained as a regression network, second, a 3D U‐Net trained as a generator of a generative adversarial network (GAN) and third, a dose transformer model which interprets the dose prediction as a sequence translation task. These models are trained to emulate the result of MC simulations. The dose depositions of a proton minibeam with a diameter of 800μm and an energy of 20–100 MeV inside a simple head phantom calculated by full Geant4 MC simulations are used as a case study for this comparison. The spatial resolution is 0.5 mm. Special attention is put on the evaluation of the generalization performance of the investigated models. Results: Dose predictions with all models are produced in the order of a second on a GPU, the 3D U‐Net models being fastest with an average of 130 ms. An investigated 3D U‐Net regression model is found to show the strongest performance with overall 61.0%±$\%\pm$0.5% of all voxels exhibiting a deviation in energy deposition prediction of less than 3% compared to full MC simulations with no spatial deviation allowed. The 3D U‐Net models are observed to show better generalization performance for target geometry variations, while the transformer‐based model shows better generalization with regard to the proton energy. Conclusions: This paper reveals that (1) all studied deep learning models are significantly faster than non‐machine learning approaches predicting the dose in the order of seconds compared to hours for MC, (2) all models provide reasonable accuracy, and (3) the regression‐trained 3D U‐Net provides the most accurate predictions. [ABSTRACT FROM AUTHOR]

There is a growing interest in developing data‐driven reduced‐order models for atmospheric and oceanic flows that are trained on data obtained either from high‐resolution simulations or satellite observations. The data‐driven models are non‐intrusive in nature and offer significant computational savings compared to large‐scale numerical models. These low‐dimensional models can be utilized to reduce the computational burden of generating forecasts and estimating model uncertainty without losing the key information needed for data assimilation (DA) to produce accurate state estimates. This paper aims at exploring an equation‐free surrogate modeling approach at the intersection of machine learning and DA in Earth system modeling. With this objective, we introduce an end‐to‐end non‐intrusive reduced‐order modeling (NIROM) framework equipped with contributions in modal decomposition, time series prediction, optimal sensor placement, and sequential DA. Specifically, we use proper orthogonal decomposition (POD) to identify the dominant structures of the flow, and a long short‐term memory network to model the dynamics of the POD modes. The NIROM is integrated within the deterministic ensemble Kalman filter (DEnKF) to incorporate sparse and noisy observations at optimal sensor locations obtained through QR pivoting. The feasibility and the benefit of the proposed framework are demonstrated for the NOAA Optimum Interpolation Sea Surface Temperature (SST) V2 data set. Our results indicate that the NIROM is stable for long‐term forecasting and can model dynamics of SST with a reasonable level of accuracy. Furthermore, the prediction accuracy of the NIROM gets improved by almost one order of magnitude by the DEnKF algorithm. Plain Language Summary: Data assimilation (DA) has been extensively used for integrating models and observations to improve weather forecasts and climate projections. The Earth system models are derived from fundamental governing equations and require the use of supercomputers to solve them numerically. One of the challenges in the DA cycle is the huge computational costs associated with these models. Therefore, new methods are warranted that can alleviate the cost of model integration within the DA cycle. Recent progress in machine learning (ML) offers an opportunity to build computationally efficient surrogate models solely from the observations and these cheap‐to‐evaluate surrogate models can accelerate the limiting bottlenecks of the DA cycle. Here, we develop a non‐intrusive low‐dimensional surrogate model that learns solely from past historical data to forecast the weekly average Sea Surface Temperature. The surrogate model is further integrated within the DA cycle for blending sparse and noisy discrete observations to improve the predictions. Overall, our results indicate that the merger of ML and DA could address the grand challenge of making use of observations to achieve robust prediction of the Earth system. Key Points: A new model is developed using deep learning to forecast the temperature in NOAA Optimum Interpolation Sea Surface Temperature V2 data setThis low‐dimensional surrogate model is integrated within data assimilation (DA) to incorporates real‐time sparse and noisy observationsThe proposed framework accelerates the DA cycle with a robust surrogate model to produce accurate state estimates [ABSTRACT FROM AUTHOR]

LINKED CONTENT: This article is linked to Verma et al papers. To view these articles, visit https://doi.org/10.1111/apt.17891 and https://doi.org/10.1111/apt.17902 [ABSTRACT FROM AUTHOR]

Since model bias and associated initialization shock are serious shortcomings that reduce prediction skills in state‐of‐the‐art decadal climate prediction efforts, we pursue a complementary machine‐learning‐based approach to climate prediction. The example problem setting we consider consists of predicting natural variability of the North Atlantic sea surface temperature on the interannual timescale in the pre‐industrial control simulation of the Community Earth System Model. While previous works have considered the use of recurrent networks such as convolutional LSTMs and reservoir computing networks in this and other similar problem settings, we currently focus on the use of feedforward convolutional networks. In particular, we find that a feedforward convolutional network with a Densenet architecture is able to outperform a convolutional LSTM in terms of predictive skill. Next, we go on to consider a probabilistic formulation of the same network based on Stein variational gradient descent and find that in addition to providing useful measures of predictive uncertainty, the probabilistic (Bayesian) version improves on its deterministic counterpart in terms of predictive skill. Finally, we characterize the reliability of the ensemble of machine learning models obtained in the probabilistic setting by using analysis tools developed in the context of ensemble numerical weather prediction. Plain Language Summary: Businesses and government agencies rely heavily on numerical predictions of climate variables such as temperature and precipitation for a wide variety of purposes ranging from integrated assessment to developing mitigation strategies to developing resilience and adaptation strategies. Developing interannual to decadal predictions using comprehensive and complex climate and earth system models, however, are computationally intensive. As such, computationally efficient and accurate surrogates of comprehensive earth system models is highly desired. Data‐driven models using advanced deep learning algorithms are promising for this purpose. This paper first considers a recently proposed convolutional network architecture to develop such a surrogate and then integrates Bayesian inference to this architecture to further assess predictive uncertainty. We show that the resulting Bayesian deep learning model not only improves prediction accuracy but also quantifies the uncertainty arising from the data and model. Key Points: Model bias and associated initialization shock are serious shortcomings that reduce prediction skill in state‐of‐the‐art decadal climate prediction effortsA complementary machine‐learning‐based approach to climate prediction is considered. Both deterministic and probabilistic machine learning approaches are examinedIn addition to providing useful measures of predictive uncertainty, Bayesian versions of deep learning models outperform their deterministic counterparts in terms of predictive skill [ABSTRACT FROM AUTHOR]

Software bug prediction aims to enhance software quality and testing efficiency by constructing predictive classification models using code properties. This enables the prompt detection of fault‐prone modules. There are several machine learning‐based software bug prediction studies, which mainly focus on single view data by disregarding the natural ordering relation among the class labels in the literature. Thus, these studies cause losing each view's own intrinsic structure and the inherent order of the labels that positively affect the prediction performance. To overcome this drawback, this study focuses on integrating ordering information and a multi‐view learning strategy. This paper proposes a novel approach multi‐view ordinal classification (MVOC), which learns from different views (complexity, coupling, cohesion, inheritance and scale) of the software dataset separately and predicts software bugs taking the inherent order of class labels (non‐buggy, less buggy and more buggy) into consideration. To demonstrate its prediction performance, the MVOC approach was executed on the 40 different real‐world software datasets using six different classification algorithms as base learners. In the experiments, the MVOC approach was compared with traditional classifiers and their multi‐view implementations in terms of precision, recall, f‐measure and accuracy rate metrics. The results indicate that the MVOC approach presents better prediction performance on average than the multi‐view‐based and traditional classifiers. It is also observed from the results that the MVOC.RF model achieved the highest classification performance with an average accuracy rate of 85.65%. [ABSTRACT FROM AUTHOR]

Summary: Nonlinearity and nonstationary data affect the prediction accuracy of network traffic. A practical solution is to use an integrated modeling method based on time‐series decomposition and an extreme learning machine (ELM). The original network traffic data series are decomposed in this paper using variational mode decomposition (VMD), and then each subdata series is subjected to phase space reconstruction (PSR). Finally, ELM trains a model for predicting network traffic. We used scalable artificial bee colony (SABC) algorithms to optimize parameter selection throughout the modeling process. However, several processes, such as VMD decomposition, PSR, and ELM training, will be repeated during iterative model optimization, resulting in a significant increase in computational complexity. This article proposes an adaptive selection mechanism for decomposing time series, which named is SABC‐VMD‐ELM prediction method with an adaptive selection (SAVE‐AS) method. From gray correlation degree and approximate entropy, an adaptive selection operator is constructed that can adaptively select the number of data sets after VMD optimization decomposition. The ELM model is trained using several subdata sequences that satisfy the conditions rather than all subdata sequences to minimize computational complexity in modeling. Two publicly available data sets of Mackey–Glass and Lorenz chaotic time series were used; in addition, actual network traffic data from the MAWI Working Group's WIDE Backbone was simulated. It is demonstrated that the prediction model can reduce computational complexity and accelerate convergence speed while maintaining the model's predictive accuracy. In three data sets, training time was reduced by 25.25%, 23.87%, and 41.36%, respectively. [ABSTRACT FROM AUTHOR]

Treatment effects are traditionally quantified in controlled experiments. However, experimental control is often achieved at the expense of representativeness. Here, we present a data‐driven reciprocal modelling framework to quantify the individual effects of environmental treatments under field conditions. The framework requires a representative survey data set describing the treatment (A or B), its responding target variable and other environmental properties that cause variability of the target within the region or population studied. A machine learning model is trained to predict the target only based on observations in group A. This model is then applied to group B, with predictions restricted to the model's space of applicability. The resulting residuals represent case‐specific effect size estimates and thus provide a quantification of treatment effects. This paper illustrates the new concept of such data‐driven reciprocal modelling to estimate spatially explicit effects of land‐use change on organic carbon stocks in European agricultural soils. For many environmental treatments, the proposed concept can provide accurate effect size estimates that are more representative than could feasibly ever be achieved with controlled experiments. [ABSTRACT FROM AUTHOR]

Background and aims: Several reports have documented risk factors for opioid use following treatment discharge, yet few have assessed sex differences, and no study has assessed risk using contemporary machine learning approaches. The goal of the present paper was to inform treatments for opioid use disorder (OUD) by exploring individual factors for each sex that are most strongly associated with opioid use following treatment. Design Secondary analysis of Global Appraisal of Individual Needs (GAIN) database with follow‐ups at 3, 6 and 12 months post‐OUD treatment discharge, exploring demographic, psychological and behavioral variables that predict post‐treatment opioid use. Setting One hundred and thity‐seven treatment sites across the United States. Participants: Adolescents (26.9%), young adults (40.8%) and adults (32.3%) in treatment for OUD. The sample (n = 1,126) was 54.9% male, 66.1% white, 20% Hispanic, 9.8% multi‐race/ethnicity, 2.8% African American and 1.3% other. Measurement Primary outcome was latency to opioid use over 1 year following treatment admission. Results: For women, regularized Cox regression indicated that greater withdrawal symptoms [hazard ratio (HR) = 1.31], younger age (HR = 0.88), prior substance use disorder (SUD) treatment (HR = 1.11) and treatment resistance (HR = 1.11) presented the largest hazard for post‐treatment opioid use, while a random survival forest identified and ranked substance use problems [variable importance (VI) = 0.007], criminal justice involvement (VI = 0.006), younger age (VI = 0.005) and greater withdrawal symptoms (VI = 0.004) as the greatest risk factors. For men, Cox regression indicated greater conduct disorder symptoms (HR = 1.34), younger age (HR = 0.76) and multiple SUDs (HR = 1.27) were most strongly associated with post‐treatment opioid use, while a random survival forests ranked younger age (VI = 0.023), greater conduct disorder symptoms (VI = 0.010), having multiple substance use disorders (VI = 0.010) and criminal justice involvement (VI = 0.006) as the greatest risk factors. Conclusion: Risk factors for relapse to opioid use following opioid use disorder treatment appear to be, for women, greater substance use problems and withdrawal symptoms and, for men, younger age and histories of conduct disorder and multiple substance use disorder. [ABSTRACT FROM AUTHOR]

Predicting the performance of thermoelectric generators (TEGs) is an essential part of designing high‐performance TEGs. However, due to the complexity of the TEG system, the existing methods are either time‐consuming or not precise enough, inconvenient for device optimization. In this paper, the deep learning (DL) method to fast and accurately get the performance of TEG devices is presented. First, the key features of a typical TEG device are captured and the training dataset is prepared based on the extracted features and finite element simulations. Next, a proper deep neural network architecture is acquired and the model is trained to converge at a low loss. Finally, the experimental data is used to validate the generalization ability of the presented model. Besides, the device optimization based on the DL solution is performed and an output power enhancement of up to 182% is achieved for the authors' sample module. The presented DL solution thus can be well applied in designing or optimizing high‐performance TEGs. Furthermore, the established framework also sheds considerable light on applying the DL approach to solve general engineering problems. [ABSTRACT FROM AUTHOR]