Search

Applying deep learning, especially time series neural network, to stock market prediction, has become one of the important applications in the quantitative finance field. However, due to the multi-correlation and volatility of the stock market, how to timely and accurately predict it has become a challenging issue. In order to cope with this challenge, a news-driven stock market index prediction model based on TrellisNet and a sentiment attention mechanism (SA-TrellisNet) is proposed. A sentiment analysis model based on CNN and LSTM is presented to obtain the sentiment index of massive news crawled from authoritative financial websites. Furthermore, a sentiment attention mechanism is designed for data fusion of stock data and news sentiment index as the input of the simple and efficient TrellisNet network for model training and prediction. The performance of our model is systematically evaluated using seven major international stock market indices including S&P500, NYSE, DJI, NASDAQ, FTSE 100, Nikkei 225 and SSE, and comparative experiments demonstrate that SA-TrellisNet is competitive to the other state-of-the-art methods in predicting stock market indices. [ABSTRACT FROM AUTHOR]

The emergence of artificial intelligence (AI) and its related capabilities has led industries to rethink the existing practices of conventional supply chain management and data analysis. Machine learning (ML), Deep Learning (DL) and their unique ability to predict future data and classify data have led to important research in the supply chain (SC) domain, particularly in identifying and prioritizing supply chain risks. This paper proposes several DL methodologies to exploit the benefit of DL, particularly to identify whether any product will be delivered late due to any unforeseen reason in a complex SC system. Four different DL architectures (Simple-LSTM, Deep-LSTM, 1D-CNN, and TCN-1DSPCNN models) are proposed to extract features, while six variant classifiers: Softmax, random trees (RT), random forest (RF), K-nearest neighbor (KNN), artificial neural network (ANN), and support vector machine (SVM), were used to classify delay or non-delay information. By seamlessly capturing intricate temporal dependencies, these DL models enhance accuracy in robustly identifying supply chain late orders. Leveraging their hierarchical feature learning, these proposed DL models excel in recognizing subtle patterns and correlations, making them ideal for classifying late orders within the supply chain. Their parallel processing prowess facilitates real-time decision support, allowing organizations to address potential delays and allocate resources effectively and proactively. Five-fold cross-validation is presented to avoid over-fitting and to prove the efficiency of the proposed DL models. The total accuracies of the six ML classifiers are 74.03, 75.81, 93.35, 87.72, 93.59, and 95.10, respectively, while the maximum accuracies obtained from four proposed DL methodologies obtained an accuracy of 97.6, 98.63, 100, 100% respectively using the SVM classifier for predicting late orders based on five-fold cross-validation. • This paper investigates a few DL approaches to extract features of SC data. • Both RNN and CNN are applied in the same model. • An improved CNN model has been proposed for feature extraction. • An online dataset is employed to validate the proposed DL architectures. [ABSTRACT FROM AUTHOR]

The research on bipedal robots with limited foot area is gaining increasing attention. To tackle the challenge of dealing with unknown disturbances in the environment, the adjustment of footstep placement plays a vital role in maintaining stable motion during bipedal walking. This paper introduces an innovative approach based on the relationship between the Divergent Component of Motion (DCM) and footstep. It utilizes a DCM prediction model to optimize the optimal speed for recovering the foothold position. The goal is to enable quick and relevant footstep selection for bipedal robots, thereby facilitating the swift recovery of robot speed. The paper provides insights into the process of designing the desired DCM for achieving an optimal average walking speed without relying on predefined footstep sequences. By establishing a state equation between the DCM and footstep placement, this approach enables the prediction of multiple footstep positions within a fixed walking cycle, thereby facilitating the desired average motion speed. Extensive numerical simulations are conducted to compare the proposed method with various conventional footstep adjustment algorithms. The results emphasize our method's ability to converge more rapidly to the target speed, even with minor step adjustments. To validate the feasibility and robustness of the algorithm, we conduct experiments on the bipedal robot BHR-B2. These experiments further confirm the algorithm's effectiveness. Given its promising performance, this algorithm holds potential for applications in legged robots with point feet. • The adjustment of the footstep is crucial for the speed recovery of bipedal walking. • A multi-step prediction footsteps adjustment algorithm based on DCM is proposed. • The desired speed of motion is mathematically related to the target DCM state. • Three adjustment algorithms were compared through simulation. • Algorithms have been applied and effectively validated on a bipedal platform. [ABSTRACT FROM AUTHOR]

Mesoscale Eddy (ME) is a widely recognized and significant oceanic phenomenon characterized by extensive energy exchange. Accurate predictions of the future geospatial distribution of ME are crucial for maritime activities. Deep learning (DL) methods for ME prediction have consistently outperformed classical mathematical models and numerical modeling approaches. However, current DL methods based on two-dimensional gridded data suffer from error accumulation issues due to their indirect prediction approach, which involves first predicting the future environmental field and then identifying the ME patterns within it. Additionally, these models ignore the potential influence of prior knowledge related to ME on prediction outcomes. To address these issues, this paper proposes a direct prediction DL network called ES-ConvGRU, which incorporates historical prior statistics. First, we conduct an analysis of ME data from the Northwest Pacific region, extracting prior statistical knowledge such as interannual variations and seasonal characteristics. Second, mining patterns for the spatiotemporal features of both ME sequence and sea surface environmental field sequence are established. Furthermore, a nonlinear approach for mapping prior statistical knowledge and a multi-step autoregressive prediction model are designed, facilitating the integration of prior knowledge with multidimensional spatiotemporal features for ME prediction. Finally, the performance and effectiveness of ES-ConvGRU are evaluated. It achieves an average pixel accuracy of 93.21%, a mean intersection over union of 81.48%, and a frequency-weighted intersection over union of 87.55% for the 7-day distribution prediction. The results show that ES-ConvGRU exhibits notable advantages compared to the existing indirect and direct prediction methods. • A direct predictive approach for mesoscale eddy distribution prediction is developed. • Prior seasonal statistics is considered for prediction. • An end-to-end deep prediction network architecture, ES-ConvGRU, is proposed. • The results show that ES-ConvGRU performs better than other predictive methods. [ABSTRACT FROM AUTHOR]

• The interval data effectively aggregated by the induced continuous aggregation operator. • The Shanghai composite index can be accurately predicted by the combination prediction model. • Thorough study of effectiveness theory of the combination prediction model has been completed. This paper proposes an interval combination prediction model for Shanghai composite index, utilizing the Theil coefficient and the induced continuous generalized ordered weighted logarithmic harmonic averaging (ICGOWLHA) operator. The effectiveness of the proposed model under specific weight conditions and the existence of its analytical solution are demonstrated. Shanghai composite index's case analysis demonstrates that, in terms of interval root mean squared error (IRMSE), interval mean absolute error (IMAE), interval mean absolute percentage error (IMAPE), and interval mean squared percentage error (IMSPE), the proposed model's predictive performance improvements over the best-performing single prediction model are 29.33%, 25.72%, 26.10%, and 28.86%, respectively. At the same time, the theoretical properties of the model are verified in the results of the case analysis, and the model's convergence is reflected in sensitivity analysis. Through extensive model comparisons, it is observed that the model proposed in this paper exhibits strong generalization, without specific limitations on data size or feature count. It demonstrates good aggregation prediction performance for interval data. Moreover, it is applicable to various fields, including finance, environment, and others. [ABSTRACT FROM AUTHOR]

Digital twins-based predictive models find their roots in smart manufacturing. However, their potential applications to control chart pattern recognition (CCPR) algorithms, which lie at the heart of advanced fault detection systems, remain underexplored. A key challenge in CCPR models arises from the intrinsic imbalance between classes, which can compromise the model's performance if left untreated. Further, existing CCPR models are often trained over simulated control chart data in which abnormal patterns are generated separately from abnormal signals; the resulting classifiers, however, perform poorly in the early detection of abnormalities in real-time production environments. To address these challenges, we develop a cost-sensitive, bi-directional long short-term memory neural network for data sequences with mixed normal and abnormal signals. We further introduce a novel adaptive weighting strategy for data generation by enforcing the rates of abnormal signals within mini-batch distributions. Our model benefits from a new bi-objective early stopping technique, which optimally balances loss minimization and G-mean maximization for model training. Finally, we introduce a novel rolling window-based metric for evaluating CCPR classifier stability. We conduct a comprehensive experimental study of our model using both simulated data and two real-world datasets collected from biomanufacturing and wafer industries. The results of our study consistently demonstrate the superiority of our proposed stopping technique over traditional methods. Our experiments further show the effectiveness of our proposed model in maintaining the classifier stability and specifying optimal process monitoring window length within the datasets. • Developed CCPR model for mixed signals. • Introduced bi-objective early stopping. • Presented a novel data weighting strategy. • Defined stability metric for CCPR. • Demonstrated model benefits with real-world data. [ABSTRACT FROM AUTHOR]

As a sparse data analysis algorithm, ensuring a reasonable model structure is an important challenge for grey models to identify the control mechanism of the uncertain system from observational data. To improve the intelligence and adaptability of the model, this study presents a synchronized optimization strategy for data prioritization and model structure for discrete polynomial grey prediction model. The proposed polynomial grey model contains two hyper-parameters: memory factor parameter and structural parameter. The memory factor is introduced into the discrete model to reconstruct the objective function of structural parameter optimization, thereby avoiding the problem of information superposition. The structural parameter is used to enhance the adaptability of grey prediction model in uncertain data analysis tasks. By employing a recursive estimation approach, an adaptive strategy for estimating model hyper-parameters is proposed, which focuses on minimizing prediction errors within the in-sample data. Additionally, a comparison is made between the proposed improved polynomial grey model and existing polynomial grey models in terms of data information mining, estimation stability, and robustness against measurement noise. The proposed model is applied to the practical engineering application of wear prediction, further validating the effectiveness of the proposed approach in non-equidistant time series prediction tasks. [ABSTRACT FROM AUTHOR]

Surface roughness is a critical parameter for quantifying the surface quality of a workpiece. Accurate surface roughness prediction can significantly influence the overall quality of the final components by reducing costs and enhancing productivity. Although various prediction methods have been explored in previous research, less attention has been given to explainable prediction methods for surface roughness. This paper proposes a surface roughness prediction model based on the Differential Evolution (DE) algorithm with ensemble learning. It aims to elucidate the intrinsic correlation between processing parameters and surface roughness values in Multi-Jet Polishing (MJP) using explainable analysis methods. The proposed Ensemble Regression with Differential Evolution (ERDE) algorithm comprises two main modules: data processing and analytics, and ensemble model prediction. These modules investigate the surface roughness of MJP from the data and model levels, respectively. To validate the effectiveness of ERDE, MJP experiments were conducted on three-dimensional printed components of 316L stainless steel. The experimental data indicated that ERDE outperforms other existing algorithms, reducing the mean absolute percentage error by approximately 42 %. [ABSTRACT FROM AUTHOR]

As excessive rain may cause numerous disasters, rainfall prediction is very crucial and the prediction should be realistic since it encourages individuals to take preventative steps. This work proposes a new rainfall prediction model by following three major phases: Preprocessing, Extraction of Features and Prediction. The Improved Data Normalization technique is carried out for preprocessing the data. Then, it extracts the technical Indicator features like Average Directional Movement (ADX), Moving Average Convergence Divergence (MACD), RSI, Welles Wilder's Smoothing Average (WWS), statistical & higher order statistical features. According to these features, prediction takes place by a hybrid model that includes the Improved DBN & LSTM methods. By fine-tuning the best weights, the training will be carried out optimally to increase the prediction model's efficacy. For this, the Brownian Motion-based Pelican Optimization Algorithm (BMPOA) is implemented. Finally, the proposed technique has contrasted over existing models concerning different metrics. [ABSTRACT FROM AUTHOR]

In steel enterprises, the long-term interval prediction of downstream gas demand plays an increasingly significant role in the scheduling decision-making of gas supply networks. Accurate long-term interval prediction remains a challenge owing to the intermittent and fluctuating characteristics of ironmaking and steelmaking. In this paper, a novel granulation-based interval prediction model (STGLUBE) is developed, which considers various data granule features and spatial–temporal correlated features. Granular evaluation indices and a corresponding training method are designed for STGLUBE training. A hybrid long-term interval prediction method integrated with a multiscale decomposition and error correction mechanism is further proposed to satisfy the actual requirements. In this method, an empirical wavelet transform is used to extract simplified modes from the original gas demand. The proposed STGLUBE is implemented in the primary mode, whereas the point predictions are performed in other modes. In addition, an error correction mechanism is introduced to enhance prediction accuracy. The effectiveness of the proposed method is verified by conducting two industrial case studies, and the results illustrate that the proposed method outperforms the existing widely used methods. [ABSTRACT FROM AUTHOR]

The growth of bike-sharing systems worldwide has led to an increased need for accurate bike-sharing flow prediction at the station level. However, previous studies mostly overlooked both substantial heterogeneity and strong correlations between the returned and rented bike flows among bike stations. This study proposes the T wo- S tream S tation-level B ike-sharing F low P rediction (TS-SBFP) framework that incorporates single-stream feature extraction, cross-stream feature interaction, global information embedding, and two-stream feature fusion for station-level bike-sharing flow prediction. In this framework, inflow and outflow data are separated into two streams and input to the prediction model respectively. First, multiscale gated convolution is applied in both streams to extract abundant temporal correlations. Second, two-stream graph attention is applied to capture rich local spatial correlations between the inflows and outflows among bike stations. Furthermore, spatiotemporal fusion network is designed to effectively and efficiently fuse the two-stream features. Additionally, to identify the predictive contributions of bike stations, global station-aware attention is designed to acquire the importance of each bike station using the squeeze-and-excitation mechanism. Finally, experiments conducted on three real-world datasets demonstrate the effectiveness of the proposed approach. In addition, two-stream attention visualization and analysis verify that the proposed approach can adaptively learn spatiotemporal interaction patterns. [ABSTRACT FROM AUTHOR]

The accurate prediction of digging load serves as a fundamental cornerstone for advancing the development of intelligent and unmanned excavators. Given the complex nonlinear dynamics of digging load, this paper proposes a novel prediction model for excavator load power based on the chaos theory and support vector regression (SVR). The presence of chaos in the dynamic digging load system is detected through phase space reconstruction. SVR is utilized for nonparametric modeling and prediction, with the reconstructed phase space capturing the essential characteristics of excavator load and serving as inputs for SVR. To optimize the hyperparameters, an improved particle swarm optimization (IPSO) algorithm is presented. Excavation experiments conducted under two typical load conditions demonstrate the superiority of the proposed chaos-based IPSO-SVR model in terms of prediction accuracy. This research lays a solid foundation for practical load prediction in industrial excavator settings. [ABSTRACT FROM AUTHOR]

Spatial prediction models hold significant application value in fields such as environmental science, economic development, and geological exploration. With advancements in deep learning technology, graph neural networks (GNNs) offer a powerful and scalable solution for spatial data modeling. In these models, each spatial location is represented as a vertex, with the explanatory variables at each point converted into vertex features, and the vertex label corresponding to the target value at that point. GNNs utilize this representation to predict vertex labels, thereby performing spatial prediction. However, the residuals of GNN predictions are found to still exhibit spatial autocorrelation, indicating that traditional GNNs only achieve suboptimal results in terms of capturing complex spatial relationships. In this paper, we propose a two-stage spatial prediction method called Location Embedded Graph Neural Networks-Residual Neural Processes (LEGNN-RNP) to address this challenge. In the first stage, we employ LEGNNs, a new framework that integrates location features into GNNs through an attention mechanism, enhancing the learning of complex spatial models by considering both spatial context and attribute features. In the second stage, we model the residuals from LEGNN predictions to further extract spatial patterns from the data. Specifically, we introduce RNP, a neural process-based approach to model the Gaussian Process (GP) distribution of residuals. The distribution parameters (i.e., mean and covariance) are parameterized by a neural network, where the mean estimates the residual and the variance quantifies the prediction uncertainty. Experiments on four datasets demonstrate that our proposed two-stage method achieves state-of-the-art results by effectively extracting spatial relationships and significantly improving spatial prediction accuracy. • Residuals of GNN regression still exhibit spatial autocorrelation. • Location Embedded GNN can learn complex spatial correlation better. • Neural embedding-based Gaussian is used to predict residuals processed by GNN. • Proposed method performs the best in spatial prediction task. [ABSTRACT FROM AUTHOR]

Road traffic crashes are a global problem and represent a major cause of mortality among children and young people worldwide. Crash prediction models (CPM) play a crucial role in identifying crash-prone areas by predicting the expected number of crashes, thus facilitating the implementation of proactive measures. Previous studies have explored various CPM based on spatial analysis or modeling approaches. Nonetheless, a substantial gap still exists in exploring CPM that simultaneously address the combination of spatial units and modeling approaches, using of heterogeneous data at both the macroscopic and microscopic levels to develop models using Multi-Criteria Decision-Making and Statistical methods. Thus, the objective of this study is to develop multiple CPM using various spatial units of analysis and employing different modeling approaches, combine them, and conduct comparisons to determine the final model that best fits the data. A total of 7 models have been developed: three at the macroscopic level, utilizing Analytic Hierarchy Process (AHP), Negative Binomial (NB), and a combination of both; one at the microscopic level using Geographic Weighted Poisson Regression (GWPR). Additionally, three combined models bring together the previous from both levels, resulting in AHP+GWPR, NB+GWPR, and AHP+NB+GWPR models. These combined and individual models have been compared, and the best-performing has been selected through a validation process. The models were calibrated using data from a case study of a highway located in the northwest of Spain for the period 2016 to 2020. Subsequently, predictions for the year 2021 were made and compared to the actual crash occurrences during that year. The results indicate that the best model is the one that combines the macroscopic and microscopic levels using the approaches AHP+NB+GWPR, which presents a Mean Absolute Deviation (MAD) of 1.55, a Mean Square Error (MSE) of 2.89, does not underestimate any segments and shows a recall of 0.88. The results obtained confirm that the combination of models at the macroscopic and microscopic levels yields the best crashes predictions. This approach offers the advantage that our model, in addition to learning from historical data using statistical methods, also fits to the importance weights assigned by knowledgeable experts on the crashes propensity of the highway in the case study area using the Analytical Hierarchical Process. [ABSTRACT FROM AUTHOR]

Accurate punctuation in written text enables unambiguous communication, minimizing the risk of misunderstandings. Conversely, faulty punctuation can confuse the intended meaning, posing challenges for the author. The existing literature offers a collection of systems and algorithms to assist users in writing tasks. However, those focusing on English tend to exhibit higher accuracy. Furthermore, most models for punctuation restoration yield results without offering insight into their decision-making processes. Therefore, this study evaluated state-of-the-art punctuation restoration models specifically for Brazilian Portuguese and incorporated the principles of explainable artificial intelligence to clarify their predictions transparently. The findings indicate that the models assessed achieved an accuracy comparable to those of their English-language counterparts. • Language Precision: Punctuation ensures clarity and proficiency in written communication. • Deep Learning Advances: BERT and T5 excel in punctuation restoration over traditional methods. • Educational Applications: Extending punctuation restoration to enhance writing skills assessment. • Model Generalizability: T5 demonstrates robust performance across diverse datasets and contexts. • Insights with XAI: Explaining model predictions aids alignment with grammatical norms. [ABSTRACT FROM AUTHOR]

Conventional machine learning methods assume that a central server holds the entire training dataset for training a global model. However, in many real-world situations, datasets are collected from individual local nodes in distributed environments. Gathering these datasets often poses challenges owing to various practical constraints such as data security and privacy issues, storage limits, and transmission costs. When data distributions differ among local nodes, local experts trained on individual local nodes exhibit poor generalization performance for new instances. In this study, we propose a training-free method for constructing an ensemble of local experts without accessing datasets. For a query instance, we apply uncertainty quantification and out-of-distribution (OOD) detection to calculate the uncertainty and OOD scores of the predictions from individual models. We then determine the weights of the models by assigning a lower weight to those with higher uncertainty and OOD scores. Using these weights, the predictions of the individual models are aggregated to obtain a final prediction. In contrast to existing methods that impose specific application requirements, the proposed method can be applied in situations in which only individual predictions from local models are available for use when making a prediction for a query instance. We evaluate the effectiveness of the proposed method using image classification benchmark datasets. • We propose a training-free method to construct an ensemble of local experts. • It eliminates the need to access training datasets. • It computes weighted combination of predictions made by individual local experts. • Weights are determined based on uncertainty and out-of-distribution scoring. [ABSTRACT FROM AUTHOR]

Deep learning models have become essential for managing time-series data across various applications in recent years. COVID-19 case data presents complex time-series patterns with high nonlinearities and dynamic fluctuations. Long short-term memory (LSTM) networks offer a suitable framework for developing prediction models to handle such complexities. However, conventional LSTM networks often have predefined, extensive structures, leading to overfitting issues and difficulties in determining the optimal number of hidden neurons. To address these challenges, we propose Regularized Self-Organizing LSTM (LSTM-ANN-RSA). This method optimizes both the structure and parameters of the network. This study introduces an adaptive learning algorithm with L2-norm regularization to adjust parameters, ensuring prediction accuracy and mitigating overfitting. Additionally, a growing strategy based on hidden neuronal sensitivity automatically determines the LSTM-ANN structure, enhancing compactness and efficiency. The proposed model's efficiency is demonstrated through comparisons with multiple deep learning models (ANN-GD, ANN-LM, ANN-RSA, GRU-ADAM and LSTM-ADAM). The results show that LSTM-ANN-RSA significantly outperformed others in predicting COVID-19 in five countries (Croatia, Greece, Italy, Poland, and Russia), with lower MAPE (2285.018, 64.4903, 205.70, 1611.19, 572.98) and higher R 2 (0.99824, 0.99726, 0.99786, 0.9962, 0.99252) values. These findings highlight the proposed model's substantial improvements in predicting the COVID-19 pandemic, enhancing the region's ability to plan for and respond to future epidemics. [ABSTRACT FROM AUTHOR]

The product design process in manufacturing involves iterative design modeling and analysis to achieve the target engineering performance, but such an iterative process is time consuming and computationally expensive. Recently, deep learning-based engineering performance prediction models have been proposed to accelerate design optimization. However, they only guarantee predictions on training data and may be inaccurate when applied to new domain data. In particular, 3D design data have complex features, which means domains with various distributions exist. Thus, the utilization of deep learning has limitations due to the heavy data collection and training burdens. We propose a bi-weighted unsupervised domain adaptation approach that considers the geometry features and engineering performance of 3D design data. The proposed method is the first to apply domain adaptation to deep learning-based engineering performance prediction to overcome the problem of insufficient industrial data. In addition, we present a source instance weighting method called the bi-weighting strategy to consider the geometry features and engineering performance of 3D design data and avoid negative transfer for accurate prediction. Domain-invariant features can be extracted through an adversarial training strategy by using hypothesis discrepancy, and a multi-output regression task can be performed with the extracted features to predict the engineering performance. The proposed model is tested on a wheel impact analysis problem to predict the magnitude of the maximum von Mises stress and the corresponding location of 3D road wheels. This mechanism can reduce the target risk for unlabeled target domains on the basis of weighted multi-source domain knowledge and can efficiently replace conventional finite element analysis. [ABSTRACT FROM AUTHOR]

Multi-Label Classification (MLC) extends standard classification in the sense that an instance might belong to multiple labels simultaneously. Many lazy approaches to MLC have been proposed so far. The majority of them, to classify an instance, use statistical estimators from the neighboring instances based on classical probability theory. In this work, we propose lazy algorithms for MLC that employ the Non-Parametric Predictive Inference Model (NPI-M) for the statistical estimators based on the neighboring instances. It is shown that our proposed lazy MLC algorithms are more suitable to tackle the class-imbalance problem that usually arises in MLC, especially when data contain label noise. This issue is corroborated via an exhaustive experimental analysis. • We propose lazy Multi-Label Classification methods based on imprecise probabilities. • They are more suitable than other methods for class-imbalance in multi-label. • Our proposed algorithms outperform other methods of the state-of-the-art. • The improvement is more notable with noise in the labels. [ABSTRACT FROM AUTHOR]

The presence of asymmetrical camber defects in reversible hot rough rolling processes significantly impacts the quality of plate shapes. Understanding and quantifying the intricate relationship between process parameters and camber is essential for developing cost-effective strategies to rectify plate shape defects. Previous research predominantly concentrated on specific couplings between process variables and camber, employing simplified mathematical models with physical mechanisms. However, these approaches may overlook critical factors inherent in the entire rough rolling process, which can substantially influence plate shape deformation. Furthermore, heavy reliance on expertise within the rolling domain may limit the capability and accuracy of coupling analysis models. In this paper, we introduce an Interpretable Prediction Model for Decoupling Camber-process Parameters (called IP-DCP), to investigate the process coupling within the entire rough rolling process. IP-DCP employs conditional modeling to integrate the intricate dependencies of process parameters during rough rolling, dynamically shaping composite features that enhance accurate camber predictions. Additionally, we introduce statistical interpretations for camber predictions using SHapley Additive exPlanation (SHAP) models. This enables a comprehensive comparison of critical process factors influencing camber and coupling analysis in various rolling scenarios, including the overall rough-rolled slab and specific steel grades such as Q335B and ST12. Extensive experiments have been conducted on actual rough roughing datasets, demonstrating the superior performance of our camber prediction model compared to baselines. Furthermore, we have conducted a quantitative analysis of their overall effect, main effect, and interaction effect. These coupling analysis results consistently align with established theoretical insights, showing their substantial significance in advancing precise process optimization strategies and cost-efficient measures for controlling camber. • First study on multi-process coupling with slab deformation. • Decouples process parameters for precise camber control in rough rolling. • IInterpretable solution elucidates camber factors and their relationships. • Coupling analysis validated by Xichang Steel's actual production data. [ABSTRACT FROM AUTHOR]

Automated Program Repair (APR) is a technique that can automatically fix software defects without manual debugging, playing a crucial role in software development and maintenance. However, the patches generated by APR still suffer from the problem of overfitting, which poses a significant threat to practical applications. Previous studies have proposed various approaches to predict the correctness of the patch to address this issue, primarily including static-based methods, dynamic-based methods, and learning-based methods. However, these methods all have their own limitations, making it difficult to accurately extract code semantic features and achieve comprehensive prediction. To address the aforementioned challenges, we propose a learning-based unsupervised classification model, A utomated P atch c O rrectness a S sessmen T based on mu L tiple p E rspectives (APOSTLE), for predicting the correctness of patches. Specifically, APOSTLE consists of three components: code vectorization component, where advanced pre-trained models are used to achieve efficient extraction of semantic features from the code; similarity and code change degree calculation component, where APOSTLE calculates the similarity and the degree of change to the code; comprehensive evaluation component, where APOSTLE conducts comprehensive evaluation, addressing the issue of prediction comprehensiveness. Experiments on a collection of 1278 patches (written by developers or generated by 32 APR tools) demonstrate that APOSTLE achieves an AUC value of 0.801, an MAP value of 0.855, and an MRR value of 0.944, outperforming the state-of-the-art approach BATS by 8.3%, 6.0%, and 9.0%, respectively. APOSTLE successfully achieves accurate extraction of code semantic features while achieving comprehensive patch correctness prediction. • APOSTLE is a learning-based unsupervised model with advanced pre-trained models. • APOSTLE solves low code semantic feature extraction during code vectorization process. • APOSTLE achieves competitive performances on large trusted patch dataset in APCA field. [ABSTRACT FROM AUTHOR]

Satellite in-orbit temperature prediction is essential in satellite reliability evaluation and health management. However, the non-stationary property of the time-series temperature data complicates forecasting, adversely affecting prediction accuracy. Moreover, conventional point estimation, which overlooks the pervasive data uncertainty due to sensor noises, diminishes the model's predictive credibility. To address these problems, this paper proposes a hybrid Monte Carlo quantile based empirical mode decomposition employing the Long Short-Term Memory method for temperature prediction and quantifying data uncertainty. The empirical mode decomposition method is adopted for tackling the non-stationary feature of satellite temperature data, greatly improving the prediction accuracy. The Monte Carlo quantile regression method is integrated with the Long Short-Term Memory network to quantify the data uncertainty, deeply enhancing the temperature prediction credibility. Two practical engineering cases are applied to validate the proposed method and address practical engineering problems. Results demonstrate that the proposed method not only provides accurate temperature predictions in comparison to other methods but also offers a reliable confidence interval at a given confidence level, facilitating timely judgments of abnormal satellite conditions. • Proposing hybrid Monte Carlo quantile based empirical mode decomposition with LSTM. • Addressing the non-stationary characteristic and quantifying data uncertainty. • Obtaining more accurate and credible predictions of satellite temperature. • Solving the practical satellite in-orbit temperature prediction issue. • Providing benefits for identifying abnormal states in advance. [ABSTRACT FROM AUTHOR]

In medical risk prediction, such as predicting heart disease, machine learning (ML) classifiers must achieve high accuracy, precision, and recall to minimize the chances of incorrect diagnoses or treatment recommendations. However, real-world datasets often have imbalanced data, which can affect classifier performance. Traditional data balancing methods can lead to overfitting and underfitting, making it difficult to identify potential health risks accurately. Early prediction of heart attacks is of paramount importance, and researchers have developed ML-based systems to address this problem. However, much of the existing ML research is based on a single dataset, often ignoring performance evaluation across multiple datasets. As the demand for interpretable ML models grows, model interpretability becomes central to revealing insights and feature effects within predictive models. To address these challenges, we present a novel data balancing technique that uses a divide-and-conquer strategy with the K -Means clustering algorithm to segment the dataset. The performance of our approach is highlighted through comparisons with established techniques, which demonstrate the superiority of our proposed method. To address the challenge of inter-dataset discrepancies, we use two different datasets. Our holistic pipeline, strengthened by the innovative balancing technique, effectively addresses performance discrepancies, culminating in a significant improvement from 81% to 90%. Furthermore, through advanced statistical analysis, it has been determined that the 95% confidence interval for the AUC metric of our method ranges from 0.8187 to 0.8411. This observation serves to underscore the consistency and reliability of our approach, demonstrating its ability to achieve high performance across a range of scenarios. Incorporating Explainable AI (XAI), we examine the feature rankings and their contributions within the best performing Random Forest model. While the domain expert feedback is consistent with the explanatory power of XAI, some differences remain. Nevertheless, a remarkable convergence in feature ranking and weighting is observed, bridging the insights from XAI tools and domain expert perspectives. • A robust data balancing technique strategically is designed to regulate the stability of classifier performance. • A structured pipeline is proposed to mitigate the problem of data discrepancies. • The structured pipeline can achieve enhanced accuracy within both single-dataset and inter-dataset frameworks. • XAI technique is used to uncover the 'inner story' hidden in black-box ML algorithms. • The XAI technique uses the domain knowledge with explainability and explores the relationship between domain knowledge and XAI tool results. [ABSTRACT FROM AUTHOR]

Accurate prediction of strip shape and reasonable allocation of actuators in the finishing mill group are essential to improving the precision of shape control. The poor prediction of the traditional mechanism model and unreasonable allocation of rolling parameters lead to various shape defects. Thus, this study developed an intelligent shape control system to enhance shape control precision, establishing a model to predict strip shape and optimize the rolling parameters for the rolling actuators in the finishing mill. A prediction model based on the deep stochastic configuration network fusion mechanism (DSCNM) model was developed to predict the hot-rolled strip shape, using ridge regression to supplement the features not provided by the mechanism model. A deep stochastic configuration network is used to approximate the errors between measured values and values computed by the mechanism model. The results show that the proposed fusion model is superior to current methods for strip shape prediction. The new model reduces prediction errors of the traditional mechanism by 62.23 %, supplying a credible foundation to optimize rolling parameters for shape control. In the optimization and allocation of rolling parameters, the rolling parameters of a finishing mill group are optimized by a sparrow search algorithm, which constructs a fitness function from the DSCNM and the target crown. A practical application of the proposed method produced a hot-rolled strip with 53.84 % fewer shape defects than the traditional method. Thus, the proposed method achieves superior quality, yield, and production cost. [ABSTRACT FROM AUTHOR]

Accurately understanding and predicting the future trajectories of pedestrians around autonomous vehicles is a major challenge. Due to the uncertainty and diversity of pedestrian trajectories, current research focuses on the multimodality of trajectory prediction. Currently, multimodal pedestrian prediction methods mainly use deep generative networks to generate mutually independent trajectories through independent sampling strategy, thus focusing on modes with a large number of samples. Moreover, deep generative networks are black-box models that are entirely data-driven and cannot effectively explain trajectory generation. In addition, current research lacks an effective measure for each trajectory. Therefore, in our paper, we propose a physics constrained pedestrian trajectory prediction with probability quantification. We divide multimodal trajectory prediction into two phases: trajectory generation phase and trajectory selection phase. During the trajectory generation phase, rather than directly predicting trajectories, we incorporate a differential constraint module to ensure that the generated trajectories adhere to pedestrian motion laws. This approach enhances the interpretability of the model prediction process. Subsequently, we generate a set of candidate trajectory proposals with specific correlations through the correlated sampling module. During the trajectory selection phase, a probability selection module is proposed to establish explicit probability distributions for the candidate trajectory proposals and to select the proposal with the highest probability as the final output. Extensive experiments on a public real-world pedestrian trajectory dataset show that our proposed model exhibits significant advantages over existing models in multimodal trajectory prediction, which not only effectively reduces the error but also mitigates mode collapse. Moreover, we provide probabilities for each trajectory to better capture the model uncertainty and provide more comprehensive information for downstream decision-making systems. • Implemented differential constraints to regulate pedestrian behavior. • Proposed a correlation sampling method to mitigate modal collapse. • We propose an efficient probabilistic measurement mechanism. [ABSTRACT FROM AUTHOR]

This study presents a novel link prediction model for the analysis and prediction of collaboration networks in the context of H2020 projects. This model incorporates various databases and expands the list of variables used in the analysis. In this study, the accuracy and interpretability of both generic and nongeneric link prediction algorithms are examined through a comprehensive analysis. Additionally, collaborative groups that exhibit stronger linkages than initially anticipated are examined, and their prevalence across different regions is evaluated. In this study, it was found that the collaboration network within H2020 has significant concentration and fragmentation levels. Moreover, several key factors that strongly influence collaboration, including prior and concurrent project involvement, earned contribution, and the economic and technological characteristics of participating businesses, were identified. This study additionally demonstrated that cooperative communities exhibit geographical concentration in western Europe, whereas nongeneric models offer improved predictive accuracy for identifying relationships. This research contributes to the field of cooperation network analysis and offers guidance to decision-makers on strategies to promote future collaborations in the context of FP. • A new link prediction model is proposed for H2020 collaboration network analysis. • Economic and corporate indicators are applied to construct a network-based database. • Generic and nongeneric methods for link prediction are applied. • The collaboration communities are identified. [ABSTRACT FROM AUTHOR]

Accurate remaining useful life (RUL) prediction holds significant importance for health management of aero-engines, ensuring safety and reducing the maintenance cost. The coupling between variable operating conditions and diverse sensor signals makes it hard to construct an accurate and stable model, which predicts precise results and simultaneously obtains condition-independent features reflecting the degradation trend. To address the problem, this paper proposes a novel approach named residual-based adversarial feature decoupling (RAFD) for RUL prediction of aero-engine which achieves decoupling both explicitly and implicitly. The paper formally defines the coupling relationships within the signals, consisting of information from normal pattern, degradation pattern, and operating conditions which are decoupled explicitly and implicitly. An operation-condition-mapping model (OCMM) is developed to explicitly decouple normal pattern, establishing the mapping between operation conditions and sensor signals in the healthy stage. Residuals serve as inputs of the prediction model, a continuous adversarial neural network specifically designed for implicit decoupling which extracts condition-independent features and predicts RUL accurately. The effectiveness of our method is validated on NASA's N-CMAPSS dataset, resulting in superior prediction performance and feature visualization when compared with other existing methods. • Signal coupling is defined and decoupled by RAFD explicitly and implicitly. • OCMM is proposed to extract residuals for explicit degradation indication. • CANN is proposed to capture implicit decoupling of operation conditions. • The superiority of the method is validated on N-CMAPSS dataset. [ABSTRACT FROM AUTHOR]

Traffic flow (TF) prediction is vital in traffic systems. Aiming at the effect of noise on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), an improved CEEMDAN based on amplitude-aware permutation entropy (AAPE), named ICEEMDAN, is proposed. Aiming at the parameter selection problems on variational mode decomposition (VMD), VMD improved by pelican optimization algorithm (POA) is proposed, called PVMD. Aiming at the parameter selection problems of bidirectional long short-term memory (BiLSTM) and kernel extreme learning machine (KELM), BiLSTM improved by sea-horse optimizer (SHO), named SHOBiLSTM, and KELM improved by nutcracker optimizer algorithm (NOA), named NOAKELM are proposed, respectively. Since the chaotic characteristics of TF, multi-factor combined TF prediction model with ICEEMDAN, PVMD, SHOBiLSTM, NOAKELM, improved entropy weight method is proposed. First, use ICEEMDAN to decompose TF and get intrinsic mode functions (IMFs). Second, refined composite multiscale dispersion entropy (RCMDE) is used to divide IMFs into the fuzzy and detail components, fuzzy components are secondarily decomposed by PVMD, IMFs and factors are multi-factor predicted by SHOBiLSTM and NOAKELM. Finally, improved entropy weight method is proposed to weight SHOBiLSTM and NOAKELM results. TF of the UK highway proves the superiority. Experiments indicate that the proposed model is better than other nine models with RMSE, MAE, MAPE and R2 of 29.5868, 22.0431, 0.0701 and 0.9947 for TF I. The outcomes suggest this proposed model precedes nine models at 99% confidence level, providing data basis and theoretical for travel plans and traffic system. [ABSTRACT FROM AUTHOR]

Few-shot extractive question answering (EQA) is a challenging task in natural language processing, whose current methods are mainly based on pretrained language models (PLMs). Data augmentation is often employed to improve the answer predictions of EQA models in few-shot settings. However, due to the differences between pretraining objectives and the EQA task, as well as embedding space alignment bottlenecks, the performance of few-shot EQA models must be improved. We propose TPKE-QA , a few-shot extractive Q uestion A nswering approach via T ask-aware P ost-training and K nowledge E nhancement, with entity-noun-oriented span selection in post-training, which can automatically generate EQA-style examples from a large-scale unlabeled corpus. By post-training based on generated examples, the gap between PLMs and the EQA task is effectively filled. To avoid embedding space alignment issues, a knowledge-enhanced sequence generation and knowledge injection approach for the EQA task enables gapless knowledge enhancement and fine-tuning on the post-trained model. In experiments, TPKE-QA achieved state-of-the-art results in most few-shot settings on the MRQA 2019 benchmark. [ABSTRACT FROM AUTHOR]

Short-term metro ridership prediction is of great significance to efficient and economic operation of Urban Rail Transit (URT) systems. With the popularity of Graph Convolution Networks (GCN) and Transformers, the recent notable metro ridership forecasting methods are GCN-based and Transformer-based models. However, existing methods face the following drawbacks. First, GCN-based models fail to effectively capture global spatial correlations which are significant for accurate prediction. Second, Transformer-based models are prone to loss temporal information due to the permutation-invariant and anti-order properties of the self-attention which they used for capturing temporal correlations. To overcome the drawbacks, we propose a novel sequence-to-sequence metro ridership prediction model, named SDT-GRU, with S tacked DT-GRU layers as both encoder and decoder. The core component of our model is DT-GRU, which integrates D ual-branch T ransformer decoder into the GRU to effectively capture global spatial correlations and temporal correlations with Transformer decoder and GRU, separately. In particular, the DT-GRU module uses one branch Transformer encoder layer to capture spatial correlations within the same timestamp, and adopts another Transformer encoder layer to implicitly capture spatio-temporal correlations among previous timestamps. Then, outputs of the two Transformer encoder layers are fed into a GRU layer to capturing spatio-temporal patterns. To evaluate the effectiveness of the proposed SDT-GRU, we conduct comprehensive experiments on three real-world metro ridership datasets from Beijing, Shanghai and Hangzhou. Experimental results demonstrate that our SDT-GRU achieves better prediction performance than the state-of-the-art baselines. [ABSTRACT FROM AUTHOR]

In this study, an ML-based online prediction strategy is developed for the thermal convexity of work rolls in hot rolling. A finite volume method simulation was developed to establish the temperature field of the work rolls and validated by measurements. A thermal convexity simulation model was established on the basis of the temperature field model to supplement the thermal convexity data. The accelerated vector that departs from the global worst solution is introduced for the PSO algorithm, and the accelerated particle swarm optimization (APSO) algorithm is proposed, which has higher optimization efficiency and better optimization effect than the pso algorithm. It is tested that APSO-RF outperforms APSO-SVM and RBF in predicting the thermal convexity of rolls, showing the best prediction accuracy and stability. In addition, the reduction of rolling has the greatest effect on the work roll convexity. The ML prediction results are very close to the industrial measurements, with small average and maximum differences. The proposed APSO-RF in-line ML method has high accuracy and efficiency and is expected to improve the quality of hot rolling in a practical way. [ABSTRACT FROM AUTHOR]

For multiple integrated energy systems with similar energy using behavior, the federated learning mechanism can build a higher precision multivariate load prediction model for each integrated energy system. However, the data of different loads in these systems often have different time scales, limited by the characteristics and frequency of data acquisition equipment. Systems often set different hyperparameters with each other to ensure prediction accuracy, which makes it difficult to use federated learning for co-training. In this paper, a multi-stage Long Short-Term Memory (LSTM) federated model based on interpolation method is proposed for multi-node multivariate load forecasting on multi-time scales. In the first stage, the load data of larger time scales are interpolated to unify all the load data to the same time scale. In the second stage, each node (i.e., each integrated energy system) uses its own data to carry out multivariate load training and establish its multivariate load forecasting model. In the third stage, based on the federated learning mechanism and fine-tuning strategy, the LSTM federated prediction of multiple loads is established by fusion training between nodes. Experimental results show that the proposed method can obtain higher precision multi-source load prediction results for multiple integrated energy systems. • A spline interpolation method is used to interpolate and supplement load data. • The local multi-load prediction model is gained by using the data of each node. • The federated learning fusion training mechanism is carried out for the model. [ABSTRACT FROM AUTHOR]

Predictive analytics aims to build machine learning models to predict behavior patterns and use predictions to guide decision-making. Predictive analytics is human involved, thus the machine learning model is preferred to be interpretable. In literature, Generalized Additive Model (GAM) is a standard for interpretability. However, due to the one-to-many and many-to-one phenomena which appear commonly in real-world scenarios, existing GAMs have limitations to serve predictive analytics in terms of both accuracy and training efficiency. In this paper, we propose FXAM (Fast and eXplainable Additive Model), a unified and fast interpretable model for predictive analytics. FXAM extends GAM's modeling capability with a unified additive model for numerical, categorical, and temporal features. FXAM conducts a novel training procedure called Three-Stage Iteration (TSI). TSI corresponds to learning over numerical, categorical, and temporal features respectively. Each stage learns a local optimum by fixing the parameters of other stages. We design joint learning over categorical features and partial learning over temporal features to achieve high accuracy and training efficiency. We prove that TSI is guaranteed to converge to the global optimum. We further propose a set of optimization techniques to speed up FXAM's training algorithm to meet the needs of interactive analysis. Thorough evaluations conducted on diverse data sets verify that FXAM significantly outperforms existing GAMs in terms of training speed, and modeling categorical and temporal features. In terms of interpretability, we compare FXAM with the typical post-hoc approach XGBoost+SHAP on two real-world scenarios, which shows the superiority of FXAM's inherent interpretability for predictive analytics. • A unified and fast interpretable model (FXAM) for predictive analytics is proposed. • FXAM extends modeling capability and efficiency of Generalized Additive Model. • FXAM conducts a novel training procedure called Three-Stage Iteration. • Joint learning and partial learning help achieve high accuracy and efficiency. • FXAM performs much better on synthetic data sets and 13 real datasets. [ABSTRACT FROM AUTHOR]

Amidst the global transition to renewable energy electricity generation, this study addresses the need for improved wind electricity generation forecasting, which is pivotal in optimizing energy strategies and meeting sustainability targets. The research introduces an innovative adaptation of the grey prediction methods, named UNAGO-GM (1, 1) , which is enhanced by the unified new-information priority accumulating generation operator (UNAGO) that emphasizes more recent data. This advanced model significantly improves the model's adaptability and accuracy in forecasting wind electricity generation within the regional hierarchy. To demonstrate the proposed model's superior forecasting efficacy, the comparative experiments are conducted in the hierarchical forecasting fashion across national, regional, and global levels by contrasting UNAGO-GM (1, 1) with prevailing benchmarks like ARIMA and LSTM. The results indicate UNAGO-GM (1, 1) ' s highest forecasting accuracy and universality, with 54.07% and 156.48% improvement rates concerning MAPE s and RMSE s across all benchmarks and forecasting levels. Furthermore, UNAGO-GM (1, 1) 's output stability is validated through the robustness tests applying Monte Carlo simulations for the intelligent algorithm stability and matrix perturbation analyses. In conclusion, the novel UNAGO-GM (1, 1) model represents a significant breakthrough in renewable energy forecasting, providing precise and adaptable predictions essential for strategic planning and policy formulation in the transition towards sustainable energy sources. [ABSTRACT FROM AUTHOR]

Building a prediction model based on both real data and expert knowledge can be challenging. We experienced this difficulty in our context of predicting fall risk factors in the elderly, where two sources of data were available: hand-reported information from medical consultations and expert information from human specialists. As is often the case, the real data were incomplete and imbalanced, and the participation of medical experts required the selection of a kind of model they can understand. This article describes the steps followed in the construction and evaluation of the model, structured as a new framework. The proposed model is a Bayesian network (BN) built from expert knowledge and learning from a real data set, guided by both the understandability of the BN graph and the performance of the model to predict risk factors. The resulting Bayesian network model includes 90 variables to evaluate 10 actionable target risk factors. It demonstrated its potential compared to other classifiers in terms of prediction performance, whilst offering a higher degree of interpretability. • Build a Bayesian Network (BN) by combining expert knowledge and a real data set. • Propose a complete framework to build an efficient and understandable BN model. • Propose a knowledge model including 90 variables and 10 target risk factors for fall. • Predict 10 risk factors for fall and compare with usual classifiers. • Improve the understandability of the BN model thanks to local graphical models. [ABSTRACT FROM AUTHOR]

In this paper, we developed a dynamic Bayesian network (DBN) model to quantify uncertainties on battlefields. The model consists of the enemy's intention prediction model and the intelligence, surveillance, and reconnaissance (ISR) reliability model quantified by using the sensor detection probability. We calibrated and validated the DBN using a historical dataset based on actual provocations by comprehensively referencing open-source intelligence, including news articles from mass media and official announcements of the Ministry of National Defense. The calibrated model was then used to predict the enemy's intention in the near future, and the accuracy of the model was 84.6%. We suggested an appropriate course of action (COA) based on the enemy's intention to expedite decision-making. We further studied the interaction between the predicted enemy's intention and the selected COA. The main results of this paper are that various information regarding enemy actions, either equality or inequality, can be incorporated into the decision-making process. [ABSTRACT FROM AUTHOR]

• Gas turbine life management monitors and assesses turbine status and conditions. • Inlet and exit air passages impact gas turbine performance, not physical attributes. • Study gap: An incomplete gas-turbine model limits big data's predictive analysis. • Big data analytics drive optimized informed integrated health monitoring decisions for gas turbines. • Data mining extends gas turbine life by forecasting faults for the future. In addition to power generation, gas turbines (GT) are crucial components in many engineering applications, including those for aircraft, ships, and other machinery that needs a prime mover. One of the main concerns was predicting a comprehensive GT model, including intake, exit, and power-generator characteristics. Hence, this study provides a comprehensive review of GT's main components, parameters, faults, and predicting models using machine learning and data mining applications. Forty-four publications between 2014 and 2023 were chosen for this review study based on the primary keywords associated with the paper title. This study focuses solely on the GT model prediction aspects, considering the significance of field distribution, applications of data mining, research goals, benefits, constraints, and contributions to aviation, the oil and gas industry, and marine propulsion fields. Based on the examined studies, ANN, ELM, and KF applications show promise for potential GT life management. The typical limitations of GT life prediction, as observed in 93.2% of the review papers, involved the absence of a comprehensive GT model that incorporates all vibration factors and the use of substantial amounts of real data for predictive modelling. Another notable finding that underscores the knowledge gap in this area is the utilization of complete GT units to monitor unit life and facilitate life management. Apparently, the development of GT life management is still riddled with several limitations, which call for more improvement. [ABSTRACT FROM AUTHOR]

A new model of utility analysis and performance prediction in crowdfunding is proposed in this paper with considering backer's behavior-related decision. To establish a connection between the backer's behavior-related decision and crowdfunding performance, the proposed model includes a modeling analysis and an empirical analysis. The modeling analysis focuses on analyzing backer's backing decision with establishing the utility-driven model, while the empirical analysis verifies the association between the utility-related features and crowdfunding performance, as well as the prediction improvement of the proposed model. Specifically, prospect theory (PT) is applied first in the modeling analysis part to analyze backer's prospect utilities, and then evidential theory (ET) is applied to aggregate the multi-source dynamic feature's utility. To validate the proposed model, and based on the results of modeling analysis, we conduct the significant test with regression model and the prediction test with several machine-learning prediction model to verify whether the proposed utility features can improve the prediction results in terms of crowdfunding performance. The empirical results show that the correlation between behavior-related factors and crowdfunding performance is significant. And the area under the curve (AUC) index of the proposed model is, on average, 7.39% higher than the baseline model on several machine-learning prediction models. In addition to providing new insights for predicting crowdfunding results, this research provides a powerful tool to assist crowdfunding platform and fundraisers in operational management with analyzing backers' backing behavior. • Modeling backers' behavior-related decisions in crowdfunding. • Empirical results validate the effect of behavior-related features on performance. • On average, the proposed model improves prediction accuracy by 7.39%. [ABSTRACT FROM AUTHOR]

Energy emission systems are often influenced by external information and policy shock. However, conventional grey models ignore the existence of the impact of multiple shock events and cumulative time-delay effects, which are crucial for achieving accurate forecasts. From a dynamic perspective, we aim to establish a multivariable grey prediction model considering the impact of multiple shock events, namely MSGTDM(1,N) model. Specifically, to accurately describe the impact of different types of external shocks on the system, three kinds of nonlinear dynamic shock functions are designed, including growth dynamic shock function, decline dynamic shock function, and slope dynamic shock function. Based on this, multi-dimensional instantaneous shock utility terms and cumulative time-delay utility terms are constructed to accurately measure shock effects. And Whale Optimization Algorithm is employed to determine the optimal parameters of the shock functions through comprehensive comparative analysis. The findings affirm the MSGTDM(1,N) model's higher predictive accuracy and validity. Consequently, the forecast results of China's carbon emissions from coal and natural gas consumption in 2022–2025 provide a reliable basis for adjusting the energy structure and implementing the "dual-carbon" policy effectively. [ABSTRACT FROM AUTHOR]

• Employed a parallel spatial-temporal convolution architecture for network-level prediction. • Implemented a multi-graph convolution module integrated with domain knowledge. • Conducted an exploration of various graph convolution structures. • Validated the approach using two real-world, large-scale datasets. Metro passenger flow prediction is a critical problem in metro transport systems. However, recent studies have either overlooked spatial information on the metro network or primarily focused on modeling spatial dependencies using only the physical topology. To achieve accurate metro passenger flow (inflow and outflow at each station of a network) prediction, this study proposes a joint prediction model that combines the multi-graph convolution network and the gated recurrent unit (GRU). In addition to exploring location topology relationships, the proposed model selects two non-Euclidean spatial dependencies in metro passenger flow prediction to design essential graph elements as part of the stacked spatial block. Three spatial relationships (adjacency, similarity, and correlation) are integrated in parallel with the GRU network. The metro passenger flow prediction framework ASC-GRU (adjacency, similarity, correlation, and gated recurrent unit) is designed to mitigate the distortion of results during the capturing of passenger flow spatial–temporal features. Finally, ASC-GRU is tested using two datasets from the Hangzhou and Shanghai metro networks in China, and the error metrics of different models are compared and analyzed to verify the effectiveness and feasibility of ASC-GRU. The test results demonstrate that the proposed model outperforms other baseline models in passenger flow prediction over long time intervals and large networks. In particular, compared with the best performance of the baselines, the average reduction is around 3%, 12% and 13% in metrics of MAPE, MAE and RMSE, respectively. [ABSTRACT FROM AUTHOR]

Information granulation theory has been widely used in short-term time-series forecasting research and holds significant weight. However, the error accumulation due to the lack of granular accuracy, along with information redundancy or deficiency in predictions, significantly affects short-term prediction accuracy. To compensate for these shortcomings, this paper proposes a double-level optimal fuzzy association rules prediction model for short-term time-series forecasting, which can strengthen the performance of information granulation in prediction. Firstly, this paper proposes a concept of breakpoints, which can accurately segment complex linear trends in time series and thus obtain a granular time series with highly accurate linear fuzzy information granules (LFIGs). Secondly, a improved distance is proposed to more accurately reflect the similarity between LFIGs by addressing counter-intuitive problems in the original distance. Theoretical analysis shows that the improved distance can effectively reduce errors in granular calculation. Then, a granule-suited fuzzy c-means algorithm is proposed for clustering LFIGs. Finally, this paper proposes a double-level optimal fuzzy association rules prediction model, which establishes the optimal rules for each cluster and selects the optimal two rules for prediction by the contribution of the clusters. The experimental results show that the prediction method effectively avoids the problems of information redundancy and information deficiency, and increases forecast accuracy. The model's exceptional performance is demonstrated through comparative analysis with existing models in experimental investigations. • A new concept of breakpoints is proposed to partition time series linear trends. • A DTW-iL1 distance is proposed to measure the similarity between granules. • DTW-iL1 Fuzzy C-Means(DFCM) is proposed for clustering granules. • Propose double-level optimal fuzzy association rules prediction model based on DFCM. [ABSTRACT FROM AUTHOR]

To simultaneously predict tool wear and surface roughness accurately during machining, this paper develops a multi-target predictive model that leverages a novel two-step feature-integration approach and Multi-kernel Gaussian process autoregressive regression (MK-GPAR). Compared with the traditional Gaussian process autoregressive regression (GPAR) models, the MK-GPAR demonstrates an enhanced capability in capturing detailed structures within heterogeneous data originating from various sources or disparate data forms. The weight coefficients of the kernel functions in MK-GPAR are typically assigned a value of 1 to minimize their influence on the predictive model. Furthermore, when compared to conventional multi-target predictive models like Gradient Boosting, Support Vector Regression (SVR), and K-Nearest Neighbors (KNN), the GPAR not only delivers higher-accuracy predictions for tool wear and surface roughness but also provides corresponding confidence intervals (CI). However, the CI of predictions for tool wear and surface roughness provided by MK-GPAR are inconsistent, which will seriously affect the reliability of the evaluation of the prediction results. The proposed two-step feature-integration approach is employed to enhance signal features effectively, reduce noise, and mitigate its adverse impact, thus enhancing the consistency and smoothness of the CI. The two-step feature-integration approach integrates principal component analysis (PCA) and orthogonal neighborhood preserving projections (ONPP) to merge force and vibration signal features obtained from the fixture's monitoring signals. Following the two-step feature-integration (first PCA, then ONPP) process, the MK-GPAR is utilized to build a multi-target predictive model for tool wear and surface roughness prediction. The impact of the proposed two-step feature-integration method on the prediction performance of the MK-GPAR is demonstrated revealed milling experiments. The experimental findings indicate that the proposed two-step feature-integration method significantly enhances the accuracy of predictions and the consistency of CI of the MK-GPAR. This research lays the foundation for implementing the multi-target predictive model in real industrial environments to predict tool wear and surface roughness. [ABSTRACT FROM AUTHOR]

High-dimensional feature selection is of particular interest to researchers. In some domains, such as microarray data, it is quite common for a group of highly correlated explanatory variables to be of equal importance for inclusion in the predictive model. This paper proposes a new hybrid feature selection approach that integrates feature screening based on Kendall's tau and Elastic Net regularized regression (K -EN). K -EN as an approach that embeds the Elastic Net, has the advantage of the grouping effect, which automatically includes all the highly correlated variables in the group. The K -EN approach offers insightful solutions to high-dimensional regression problems and improves Elastic Net performance since the screening phase is preceded by a step that further reduces the number of explanatory variables by removing those that disagree with the target based on Kendall's tau. The use of Kendall's tau further enhances Elastic Net performance, as it is robust enough to handle heavy-tailed distributions, non-parametric models, outliers, and non-normal data with greater ease. K -EN is therefore a time-saving approach. The proposed algorithm is evaluated on four simulation scenarios and four publicly available datasets, including riboflavin, eyedata, Longley, and Boston Housing, and achieves 0.2528, 0.0098, 0.1007, and 0.4121 respectively as the Mean Squared Error (MSE). K -EN's MSEs are the best compared to those achieved by the state-of-the-art approaches reviewed in this paper. In addition, K -EN selects up to 100% of relevant features when run on simulated data. [ABSTRACT FROM AUTHOR]

To alleviate the concept drift of an offline prediction model for the furnace temperature in a municipal solid waste incineration (MSWI) process caused by changes in working conditions, a prediction method for the furnace temperature based on a knowledge transfer online stochastic configuration network is proposed in this work. The proposed method includes offline learning and online learning. First, a stochastic configuration network is utilized to construct the offline furnace temperature prediction model, and a knowledge transfer method is employed to update the model under new operating conditions in the offline learning stage. Then, the updated model is used as the initial state of online modeling. Second, the recursive solution of the model output weights is presented to adapt to the dynamic change in incineration conditions, and a direction forgetting mechanism is utilized to enhance the result of the prediction model for the furnace temperature under nonpersistence of excitation in the online stage. Finally, to further verify the proposed online modeling method, the real historical data of an MSWI plant are utilized to finish the comparative experiments. The experimental results with the other methods show that the proposed prediction method for the furnace temperature presents a smaller error. Hence, the proposed method can reduce the influence of working conditions on the accuracy of furnace temperature prediction models. [ABSTRACT FROM AUTHOR]

This study proposes a new machine learning (ML) framework, which mainly includes dataset cleaning processing as well as performance predicting, for property prediction of ultra-high performance concrete (UHPC). Firstly, the missing data in original dataset is interpolated and discussed by visualization results. Then, the existing outliers in the completed dataset are found out to improve the quality of dataset. Meanwhile, by analyzing the influence of key parameter, it not only clarifies the influence of dataset quality on model prediction results, but also proves the necessity of anomaly detection, with R2 increasing 15 % and RMSE decreasing 37 %. Finally, the chosen model is trained and further optimized by hyperparameter optimization, in which the loss function is significantly reduced by 68.82 % for training data (R2 > 0.95) and 84.36 % for testing data (R2 > 0.94). Overall, this framework can effectively improve the accuracy and generalization of UHPC predictive models, which is also suitable for other types of concrete materials. [ABSTRACT FROM AUTHOR]

Seepage pressure data analysis is considered one of the most direct ways to evaluate seepage safety in dam engineering. Therefore, accurate and consistent prediction of seepage pressure is essential. However, satisfactory predictive performance is rugged due to the irregular, nonlinear, and non-stationary nature of seepage pressure monitoring data. This paper proposes a combined optimization prediction model for seepage pressure prediction to resolve the deficiency. The model is coupled with a data decomposition technology, multi-machine learning fusion technology, adaptive variable weight combined strategy, and an improved optimization algorithm. The interaction between high and low frequency components is eliminated by improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) technique to achieve the smoothing of non-stationary data. Multi-machine learning models are integrated by the combined strategy to compensate for the shortcomings of single models. Combined weights are adaptively changed by the kernel extreme learning machine (KELM) variable weighting strategy to overcome the limitations of traditional fixed weights. For the challenge of determining KELM parameters, the artificial electric field algorithm is improved to optimize KELM parameters to accelerate the determination of adaptive weights. The application of the proposed model to an earth-rock dam demonstrates enhanced predictive accuracy of seepage pressure by implementing strategies. The model is an effective tool for the seepage pressure prediction of dams and is conducive to dam operation and maintenance management. [ABSTRACT FROM AUTHOR]

Traffic flow prediction is of paramount importance in the field of spatio-temporal forecasting. In recent years, research efforts have primarily been directed towards developing intricate graph convolutional networks (GCNs) to capture spatial complexities. However, this has inadvertently led to the neglect of the inherent temporal correlations in traffic prediction, as well as the heterogeneity of graph structures. As a result, existing models show limited efficacy when dealing with the complex nature of traffic data. To address this issue, this paper introduces a novel traffic prediction model: the Fourier-enhanced heterogeneous graph convolution attention recurrent network (FEHGCARN). This model integrates historical information and incorporates a graph convolution attention recurrent unit (GCARU), meticulously engineered to effectively capture spatio-temporal dependencies. Additionally, it features a Fourier-enhanced heterogeneous graph learning module, which facilitates the acquisition of complex relationships among nodes in the frequency domain. Notably, this memory network excels at recognizing abrupt traffic conditions. To validate our approach, we conducted comprehensive comparisons using three authentic datasets and benchmarked our model against six state-of-the-art baseline methods. The experimental results unequivocally demonstrate the superior performance of our model across all evaluation metrics. • Fourier-enhanced heterogeneous graph convolution attention recurrent network. • Capturing spatio-temporal dependencies in graph convolution attention units. • The novel approach is the Fourier-enhanced heterogeneous graph learner. • Our model surpasses baselines, achieving state-of-the-art results in experiments. [ABSTRACT FROM AUTHOR]

Multivariate time series data with multiple points is a kind of spatio-temporal data, and in recent years, spatio-temporal graph neural networks (STGNNs) have become one of the most effective methods for spatio-temporal data prediction. By leveraging graph neural networks and sequence models, STGNNs simultaneously capture the hidden temporal and spatial patterns in spatio-temporal data. However, there exist two key limitations: (1) Most existing STGNNs whether based on static graphs or dynamic graphs, fail to introduce time periodicity into graph structure learning and ignore the importance of multi-relational graph structure learning. (2) Most multivariate time series prediction models with multiple points ignore the relation between multiple attributes. To this end, we propose a novel framework called the Dynamic Multi-fusion Spatio-temporal Graph Neural Network (DMF-STNet), where multi-fusion embodies two meanings: spatio-temporal fusion and multi-attribute fusion, as well as multi-graph fusion and multi-layer fusion. The dynamic periodic graph is proposed to mine time periodicity that is hidden in graph structures, and we further offer multi-relational graph structure learning methods to capture static and dynamic relations. Moreover, we introduce a dual-stage feature aggregation method to aggregate hidden features from both primary and auxiliary attributes. Extensive experiments on three real-world datasets demonstrate the efficacy of our proposed DMF-STNet, and the visualization study proves the interpretability of learned weights. • Modeling multivariate time series with multi-point. • Design a graph structure learning method with multi-relation. • Aggregating features through the dual-stage feature aggregation method. • Experiment results demonstrate the effectiveness of the proposed DMF-STNet. • Visualization study further proves the interpretability of learned weights. [ABSTRACT FROM AUTHOR]

In the context of today's energy shortage, wind energy plays a crucial role as one of the most widely used renewable energy sources. However, in order to fully utilize the potential of wind energy, it is necessary to build a precise and reliable wind-speed prediction system. Conventional wind-speed prediction models frequently rely on a solitary model for forecasting, and sometimes these models cannot accurately fit the nonlinear characteristics of the data, resulting in poor prediction performance. Therefore, to address this deficiency, a new wind-speed point-interval prediction system is proposed in this study. The fuzzy information granulation technology is adopted in this system to reduce the dimension of data and solve the problem of redundancy in wind-speed data. In addition, this paper uses the multi-objective dragonfly algorithm and combines sub-models using the multi-nonlinear weight strategy to address the issue of inadequate precision exhibited by an individual prediction model. In order to verify the effectiveness of this wind-speed prediction system, the Penglai wind farm in Shandong Province is used as the dataset to construct the model, and several reference models are selected to conduct point prediction and interval prediction comparison experiments respectively. The experimental results show that the combined prediction system has the best prediction performance, with MAPE values of 14.16%, 15.02% and 15.74%, respectively. Compared with other benchmark models, the average improvement rates for the multi-step prediction were 14.97%, 16.48% and 19.11%, respectively. [ABSTRACT FROM AUTHOR]