Search

Background: Due to the high cost of data collection for magnetization detection of media, the sample size is limited, it is not suitable to use deep learning method to predict its change trend. The prediction of physical and chemical properties of magnetized water and fertilizer (PCPMWF) by meta-learning can help to explore the effects of magnetized water and fertilizer irrigation on crops. Method: In this article, we propose a meta-learning optimization model based on the meta-learner LSTM in the field of regression prediction of PCPMWF. In meta-learning, LSTM is used to replace MAML's gradient descent optimizer for regression tasks, enables the meta-learner to learn the update rules of the LSTM, and apply it to update the parameters of the model. The proposed method is compared with the experimental results of MAML and LSTM to verify the feasibility and correctness. Results: The average absolute percentage error of the meta-learning optimization model of meta-learner LSTM is reduced by 0.37% compared with the MAML model, and by 4.16% compared with the LSTM model. The loss value of the meta-learning optimization model in the iterative process drops the fastest and steadily compared to the MAML model and the LSTM model. In cross-domain experiments, the average accuracy of the meta-learning optimized model can still reach 0.833. Conclusions: In the case of few sample, the proposed model is superior to the traditional LSTM model and the basic MAML model. And in the training of cross-domain datasets, this model performs best. [ABSTRACT FROM AUTHOR]

In order to solve the problem of deterioration of the generalization ability caused by support vector machine (SVM), this paper proposes a regression prediction method based on SVM ensemble learning. The grid search method is used to optimize the modeling parameters of an SVM-based predictor. An AdaBoost method is used to integrate multiple SVM-based predictors, and a regression prediction model based on SVM ensemble learning is constructed. Using the database collected by a hospital taken as the research object, the executing time prediction of outpatient examination scheduling was tested and compared with the experimental results of the SVM predictor. The results show that the ensemble learning algorithm can effectively reduce the computational complexity brought in by training all samples altogether and improve the prediction accuracy. The prediction instability and low precision of the sampling-based standard SVM predictor are also solved effectively. [ABSTRACT FROM AUTHOR]

When the electronic nose (E-nose) is used to predict the concentration of mixed gas, the traditional regression prediction algorithm may lead to unsatisfactory prediction results and long training time. In order to improve the accuracy of regression prediction and reduce the training time, we proposed a regression prediction algorithm based on broad learning system (BLS) to predict the concentration of mixed gas. To further improve the accuracy of model predictions, we optimize the various parameters existing in the model to improve the performance of the model. Then, we change the initial random mapping weight assignment method of the model to further improve the data processing ability of the model, and the improved model is called GBLS. In the data processing experiment, we use the mixed gas of methane and ethylene as the test gas to test the GBLS model proposed in this article. We have compared GBLS with other existing methods including back propagation neural networks (BPNN), least squares support vector machines (LSSVM), extremely learning machine (ELM), linear regression (LR). Experimental results show that the proposed GBLS outperforms the other methods. [ABSTRACT FROM AUTHOR]

Machine learning is widely used for predicting the compressive strength of concrete. However, the machine learning modeling process relies on expert experience. Automated machine learning (AutoML) aims to automatically select optimal data preprocessing methods, feature preprocessing methods, machine learning algorithms, and hyperparameters according to the datasets used, to obtain high-precision prediction models. However, the effectiveness of modeling concrete compressive strength using AutoML has not been verified. This study attempts to fill the above research gap. We construct a database comprising four different types of concrete datasets and compare one AutoML algorithm (Auto-Sklearn) against five ML algorithms. The results show that Auto-Sklearn can automatically build an accurate concrete compressive strength prediction model without relying on expert experience. In addition, Auto-Sklearn achieves the highest accuracy for all four datasets, with an average R2 of 0.953; the average R2 values of the ML models with tuned hyperparameters range from 0.909 to 0.943. This study verifies for the first time the feasibility of AutoML for concrete compressive strength prediction, to allow concrete engineers to easily build accurate concrete compressive strength prediction models without relying on a large amount of ML modeling experience.

Forecasting the trend of coal flow in a short-term is the precondition to realize matching of belt speed and traffic volume. However, the existing short-term prediction methods of coal flow on conveyor belt have the problems of insufficient real-time performance and low precision. For the problems, a short-term prediction method of coal flow on conveyor belt based on support vector machine was proposed. Firstly, the method uses support vector machine algorithm to select the real-time coal flow as dependent variable and statistical data time as independent variable, and then normalizes actual collected coal flow data, uses cross-validation method to select the best parameters. It also uses the best parameters to train the support vector machine to fit ideal short-term prediction curves of coal flow. Finally, the method analyzes fitting degree of the coal flow prediction curves and the original data curves by further comparing the prediction parameters such as mean square error and correlation coefficient to obtain the best prediction curve. The Matlab simulation results show that the method can predict coal flow on conveyor belt in a short time, and the deviation between the predicted data and the true value is small, the mean square error is 0.000 152 563, and the correlation coefficient is 99.784 8%.

The observation errors of independent variables in model prediction are typically neglected in traditional regression models, which leads to the decreased prediction accuracy. In this paper, a new regression prediction method that is based on the partial errors-in-variables (partial EIV) model is proposed, which takes into account the observation errors of all variables. It observes that each element of the coefficient matrix is an expression or a function. The partial EIV model is transformed into a linear model and constructed in the form of indirect adjustment for iterative solution. By considering the errors in the independent variables when predicting the corresponding dependent variables with the partial EIV model, the proposed method can achieve higher prediction accuracy. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Frontiers of Computer Science & Technology is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of China Petroleum Machinery is the property of China Petroleum Machinery Editorial Department and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Journal of Mine Automation is the property of Industry & Mine Automation Editorial Department and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

To facilitate accurate control of the output force at the arm end during live working,a prediction method for arm end force based on surface electromyography (sEMG) and support vector machine regression (SVR) is proposed.By forcing the handle at the end of the manipulator to provide the reciprocating push and pull movement,the data of the force sensor at handle Fare recorded.Simultaneously,the EMG signals of the arm are synchronously collected by three sets of EMG sensors.The feature extraction from the EMG signal and F are combined as sample set S inthe sample collection.Half of the sample data is randomly selected as the training set for different machine learning methods,and trained using the back propagation,generalized regression,and SVR neural networks.Finally,the trained neural network is used to predict force F in the whole sample set,and the prediction effect of the model is obtained by means of the root mean square error (RMSE) and correlation coefficient.The results show that the prediction effect of the SVR neural network is better,the RMSE is 3.074 0,and the correlation coefficient R is 0.951 7. [ABSTRACT FROM AUTHOR]

In order to correct the error data contained in the sequential data collected by sensor nodes, in the relatively harsh deployment environment of sensor nodes, limited sensor nodes make WSN (Wireless Sensor Network) have monitoring blind areas. Based on Kriging interpolation method and natural neighbourhood interpolation algorithm, the problems of data specification and spatial interpolation in WSN network are solved. The research results show that the algorithm divides irregular meshes into regions to be interpolated, and then the k-hop neighbour nodes of the prediction points are determined as the parameter set of Kriging interpolation. Finally, the Kriging coefficients are solved, and the prediction data of the points to be interpolated are calculated according to the Voronoi area and the observation values of the neighbour nodes, thus the value of the prediction points is calculated. It can be seen that this saves time and space greatly, and simulation experiments show that the natural neighbourhood interpolation results are closer to the real value. [ABSTRACT FROM AUTHOR]

A new method relying on machine learning and resistivity to predict concentrations of petroleum hydrocarbon pollution in soil was proposed as a means of investigation and monitoring. Currently, determining pollutant concentrations in soil is primarily achieved through costly sampling and testing of numerous borehole samples, which carries the risk of further contamination by penetrating the aquifer. Additionally, conventional petroleum hydrocarbon geophysical surveys struggle to establish a correlation between survey results and pollutant concentration. To overcome these limitations, three machine learning models (KNN, RF, and XGBOOST) were combined with the geoelectrical method to predict petroleum hydrocarbon concentrations in the source area. The results demonstrate that the resistivity-based prediction method utilizing machine learning is effective, as validated by R-squared values of 0.91 and 0.94 for the test and validation sets, respectively, and a root mean squared error of 0.19. Furthermore, this study confirmed the feasibility of the approach using actual site data, along with a discussion of its advantages and limitations, establishing it as an inexpensive option to investigate and monitor changes in petroleum hydrocarbon concentration in soil. • The prediction of soil petroleum hydrocarbon concentration is achieved by machine learning and resistivity tomography method. • Building the mapping relationship between soil petroleum hydrocarbon pollution concentration and resistivity. • External validation data verify feasibility of predicting soil petroleum hydrocarbon contamination concentrations based on resistivity. [ABSTRACT FROM AUTHOR]

The implementation of neural network regression prediction based on digital circuits is one of the challenging problems in the field of machine learning and cognitive recognition, and it is also an effective way to relieve the pressure of the Internet in the era of intelligence. As a nonlinear network, the stochastic configuration network (SCN) is considered to be an effective method for regression prediction due to its good performance in learning and generalization. Therefore, in this paper, we adapt the SCN to regression analysis, and design and verify the field programmable gate array (FPGA) framework to implement SCN model for the first time. In addition, in order to improve the performance of the SCN model based on the FPGA, the implementation of the nonlinear activation function on the FPGA is optimized, which effectively improves the prediction accuracy while considering the utilization rate of hardware resources. Experimental results based on the simulation data set and the real data set prove that the proposed FPGA framework successfully implements the SCN regression prediction model, and the improved SCN model has higher accuracy and a more stable performance. Compared with the extreme learning machine (ELM), the prediction performance of the proposed SCN implementation model based on the FPGA for the simulation data set and the real data set is improved by 56.37% and 17.35%, respectively.

Nanocellulose-modified presspaper is a promising solution to achieve cellulose insulation with better performance, reducing the risk of electrical insulation failures of a converter transformer. Predicting the dielectric properties will help to further design and improvement of presspaper. In this paper, a multivariable method was adopted to determine the effect of softwood fiber on the macroscopic performance of presspaper. Based on the parameters selected using the optimum subset method, a multiple linear regression was built to model the relationship between the fiber properties and insulating performance of presspaper. The results show that the fiber width and crystallinity had an obvious influence on the mechanical properties of presspaper, and fiber length, fines, lignin, and nanocellulose had a significant impact on the breakdown properties. The proposed models exhibit a prediction accuracy of higher than 90% when verified with the experimental results. Finally, the effect of nanocellulose on the breakdown strength of presspaper was taken into account and new models were derived. [ABSTRACT FROM AUTHOR]

Based on the current concentration levels of gas components in the air, indoor air quality may be assessed. Gas information of mixed gas with various concentrations can be acquired and assessed using an electronic nose (E-nose). For mixed gas of varied concentrations, the E-nose's prediction accuracy using the conventional regression prediction algorithm is unsatisfactory. We propose a two-channel temporal convolutional network based on a model temporal convolutional network (TCN) suitable for time series processing that was utilized to increase the precision of the regression prediction of mixed gas concentration on the E-nose. In addition to improve the structure of the model, we also analyse its activation function. In the experiment, we use two types of gas mixture data with different concentrations, methane-ethylene and carbon monoxide(CO)-ethylene, in our experiments. It is compared with some baseline models including long short-term memory (LSTM), gated recurrent unit (GRU) and generic temporal convolutional network (TCN). In our experiments, we show that the two-channel TCN architecture is superior to both the models mentioned above and the generic TCN architecture. • An electronic nose is used to make regression prediction of mixed gas concentration. • Successfully determined CO-ethylene and methane-ethylene concentrations. • A new deep learning model called two-channel TCN is proposed which is based on TCN. • The proposed two-channel TCN achieves outstanding regression prediction performance. [ABSTRACT FROM AUTHOR]

It is significant to predict accurate protein melting temperature in protein engineering and drug design. This paper proposed a novel weight-based dimensionality reduction algorithm, and applied it to obtain the input features of MLP model by using combination with global and sequential features as preliminary features. On blind test sets, the PCC value of predicted and experimental melting temperatures increased from 0. 77 to 0.8, and RMSE value decreased from 0.17 to 0.16. The classification accuracy of predicted protein melting temperatures by this algorithm was significantly improved over the up-to-date method. [ABSTRACT FROM AUTHOR]

• Failure to check statistical significance leads to errors, especially with respect to physical boundary conditions. • C Appropriate use of regression yields better predictions. • Boundary condition checking is essential for fitted functions accidental matching of prediction with expectation is misleading. • Detailed analysis has provided a rational prediction and revealed potential problems with literature data. [Display omitted] The use of fitted functions without reference to boundary conditions, physical implications, or the statistical contribution of all terms is dangerous. A crystallographic example of such errors is considered (ABiO 3), problems identified, and a better approach suggested. [ABSTRACT FROM AUTHOR]

The leaching risk of heavy metals in soil has a large spatial variability on a regional scale. Taking the Chang-Zhu-Tan area as the research object, this work studied the distribution and influencing factors of available contents and solid-solution partition coefficient ( K d ) of Cd and Pb in soil with land uses and clarified the environmental risk of heavy metals in soil based on K d values measured by CaCl 2 (soil-to-water ratio, 1:0.5). The results showed that the contents of available Cd and Pb in soil followed the order of forest land>suburban farmland>urban green space>industrial green space. The average K d of Cd in soil was 449.79 L·kg -1 , and that of Pb was 27604.07 L·kg -1 , indicating that the mobility of Cd in the soil was significantly higher than that of Pb. The K d values of forest soil were significantly lower than that in the other land uses. The K d values were mainly affected by soil pH and the total content of heavy metals in soil. Adopting the available content of heavy metals measured by CaCl 2 (soil-to-water ratio, 1:10) as a dependent variable, the multiple regressions effectively predicted the K d values of Cd and Pb in soil, with R 2 values of 84.2% and 67.6%, respectively. The environmental risk assessment indicated that the leaching risk in 93.8%-96.1% of the sampling sites could be ignored, whereas a few sampling sites near factories with low pH may pose a risk to the groundwater environment. The mobility of heavy metals in soil and the distribution of pollution sources determined the leaching risk of heavy metals. The results provide a method and theoretical support for preventing the environmental risk of heavy metals in soil on a regional scale.

• The physicochemical processes and effects of external factors are represented by MLP and RF models. • The error of prediction using MLP and RF with 204 CPT experimental data is less than 10%. • RF performs better in generalization than MLP. • Machine learning models are greatly dependent on both data size and CPT distribution. • Machine Learning is demonstrated as a novel approach for effective prediction of coal self-ignition tendency. Self-ignition of coal emits hazardous particles and toxic gases, polluting environment and threatening people's health. Prediction of self-ignition tendency of coal is of great significance to prevent hazards of coal self-ignition. However, it is very challenging to forecast the self-ignition tendacy of coal, because of complex physicochemical processes and highly nonlinear correlation between factors and self-ignition tendency. In this work, machine learning methods (Multilayer Perceptron (MLP) and Random Forest (RF)) are used to represent the complex physicochemical processes and effects of external factors. The regression prediction models with regarding to crossing point temperature (CPT) and 13 input features are established. The dependence of input features is examined using the feature engineering. Two hundreds and four CPT samples are collected, in which 142 (70%) samples and 62 (30%) samples are divided as training data and testing data, respectively. Results show that the accuracy of both MLP and RF predicted CPTs in the testing data reaches 90%, which proves good predictability of machine-learning based models with several hundreds of samples. This work improves prediction of the self-ignition tendency of coal impacted by complex physicochemical properties and a variety of external factors. It may help to predict other fuels susceptible to self-ignition e.g., oil shale and biomass fuels. [ABSTRACT FROM AUTHOR]

• The fundamental matrix (F-Matrix) estimation is a classic problem in computer vision. The traditional algorithms require high-precision correspondences. However, correspondences in biplanar radiographs (BR) are difficult to match accurately. We propose an end-to-end network (FM-Net) to estimate the F-Matrix directly from BR. There is no publicly available dataset of BR. We produce the dataset in this paper to train and test the proposed network. Four metrics, MSE (Mean Square Error), R2 (Calculating R-squared), ECSV (Square Value of Extreme Constraint), and ECAV (Absolute Value of Extreme Constraint) are used to measure the performance of the approaches. The best ECSV and ECAV values we obtained on the datasets were 0.20 and 0.43, respectively. Compared with other methods, the estimation accuracy of FM-Net is improved by more than 53.53%. The results of experiments demonstrate that the proposed FM-Net can estimate the F-Matrix successfully. It outperforms the classical algorithms and other deep learning-based methods. • The main highlights of our work are as follows. • Siamese convolutional networks with shared weights are designed for extracting discrepancy features. • Network layers are designed to constrain the mathematical properties of the F-Matrix to improve the accuracy of the F-Matrix estimation. • Input angle information into the network to improve the F-Matrix estimation accuracy. • Input the position information in the feature map into the network to improve the F-Matrix estimation accuracy. • Introduce Convolutional Block Attention Module (CBAM) to improve the F-Matrix estimation accuracy. And the CBAM structure is adjusted to reduce the number of its parameters. • Four publicly available biplanar radiographs datasets are created. The fundamental matrix estimation is a classic problem in computer vision. The traditional algorithms require high-precision correspondences. However, correspondences in biplanar radiographs are difficult to match accurately. We propose an end-to-end network to estimate the F-Matrix directly from BR, which includes feature extraction and regression prediction. There is no publicly available dataset of biplanar radiographs. We produce the dataset in this paper to train and test the proposed network. Four metrics, Mean Square Error, Calculating R-squared, Square Value of Extreme Constraint, and Absolute Value of Extreme Constraint are used to measure the performance of the approaches. The best Square Value of Extreme Constraint and Absolute Value of Extreme Constraint values we obtained on the datasets were 0.20 and 0.43, respectively. Compared with other methods, the estimation accuracy of FM-Net is improved by more than 53.53%. The results of experiments demonstrate that the proposed network can estimate the fundamental matrix successfully. It outperforms the classical algorithms and other deep learning-based methods. [ABSTRACT FROM AUTHOR]

Flowchart of our method. [Display omitted] • Collected aflatoxin B1 spectrum information on pixel-level. • Proposed new method of aflatoxin detection inspired by quantitative remote sensing. • Achieved the prediction of aflatoxin B1 content based on deep learning. Aflatoxin is a highly toxic substance dispersed in peanuts, which seriously harms the health of humans and animals. In this paper, we propose a new method for aflatoxin B1(AFB1) detection inspired by quantitative remote sensing. Firstly, we obtained the relative content of AFB1 at the sub-pixel level by subpixel decomposition (endmember extraction, nonnegative matrix decomposition). Then we modified the transfer learning models (LeNet5, AlexNet, VGG16, and ResNet18) to construct a deep learning regression network for quantitative detection of AFB1. There are 67,178 pixels used for training and 67,164 pixels used for testing. After subpixel decomposition, each aflatoxin pixel was determined to contain content, and each pixel had 400 hyperspectral values (415–799 nm). The experimental results showed that, among the four models, the modified ResNet18 model achieved the best effect, with R2 of 0.8898, RMSE of 0.0138, and RPD of 2.8851. Here, we implemented a sub-pixel model for quantitative AFB1 detection and proposed a regression method based on deep learning. Meanwhile, the modified convolution classification model has high predictive ability and robustness. This method provides a new scheme in designing the sorting machine and has practical value. [ABSTRACT FROM AUTHOR]

Abstract: Linz Donawitz converter gas (LDG) is the significant secondary energy resource that plays a crucial role in the energy system of steel industry. Since the real-time prediction for the gas tank level of LDG system is the foundation of energy balance scheduling that directly affects the energy costs of enterprise, more and more attentions has been paid to this issue. In this study, taking the LDG system of Ma’anshan Steel Co., Ltd, China into account, a multi-output least square support vector regressor is proposed, which considers not only the single fitting error of each tank level but also the combined one. Then, a prediction model for the multi-tank LDG system is derived, and a particle swarm optimization is designed to determine the parameters of this model for the sake of improving the prediction accuracy. The experimental results based on the real data from the plant demonstrate that the proposed method is effective to the practical application. [Copyright &y& Elsevier]

• The CSD can operate continuously even though the key sensor fails during construction. • All CSD sensors have a digital twin backup. • The Filter method computes less expensive than the Wrapper and Embedded methods with the same accuracy in the CSD data set. • A sensor association digital twin network is established based on the stacking generalization model. In practice, the construction environment of the cutter suction dredger (CSD) is inclement, which results in a high frequency of the key parameter sensor failure. If the key parameter data is missing or false, it will affect the continuity of the construction. This paper proposed a sensor network regression prediction method based on the "Digital Twin" to establish a correlated model between the key sensor and other highly reliable sensors in the CSD construction. The stacking model is trained by learning the CSD construction data that can synchronously calculate any key parameters when the dredger is running. The proposed method is validated on the "Changshi 12" CSD construction case. The results indicate that the method has high prediction accuracy and computes less expensively. Thus, the proposed method could better solve the problem of construction discontinuity caused by key sensor failure. [ABSTRACT FROM AUTHOR]

The distribution of the rotor slot wedge eddy current loss in large nuclear power turbo generators is complex and is influenced by many factors. Excessive eddy current loss leads to severe rotor heating, potentially leading to thermal accidents; therefore, the design precision of large generators must be improved. In this paper, a Fuzzy C-Means Deep Gaussian Process Regression (FCM-DGPR) method is proposed to predict the eddy current loss of a large generator in order to solve the problem of the insufficient accuracy of deep Gaussian process regression (DGPR) with increasing number of the data samples. First, the original dataset is obtained by the finite element method (FEM) and then normalized to construct the samples of the eddy current loss of a large nuclear power generator. Second, the training set is automatically clustered into different subsets by the fuzzy c-means algorithm, and each subset is used to train the DGPR model to obtain different sub models. The membership degree of each data point in the test set is calculated and used to evaluate the sub model of the data. Then, the sub model is used to predict the eddy current loss. Finally, the result is obtained by concatenating the results of each sub model. The results show that the goodness of fit (R 2) is 0.9809, the root mean square error (RMSE) is 0.0271, the prediction error is small, and the model exhibits good prediction performance. Further experimental results show that the FCM-DGPR method is superior to the existing DGPR models and other models and is more suitable for predicting the eddy current loss of large generators. [Display omitted] • 10,000 samples of a nuclear generator eddy current loss are obtained by FEM. • A novel fuzzy c-means deep Gaussian process regression model is established. • The FCM-DGPR method has good performance on predicting the eddy current loss. • Final result gives score evaluation with the highest R 2 of 0.9809. • The FCM-DGPR method shows higher predictive ability than other methods. [ABSTRACT FROM AUTHOR]

Today artificial intelligence is a major utility technology that can help industry and research in many fields, including solar thermal energy for water desalination purposes. Hence, the main aim of the paper focus on utilizing an artificial intelligent regression model to predict the efficiency of water desalination, which depends on solar thermal energy. The contribution helps manufacturers and research determine the productivity of the designed desalination system in the beta environment before realizing the system. The used desalination system merges and hybrids in our experimental modelling solar still and air humidification and dehumidification. Regards the prediction of performance and productivity, a wide range of regression algorithms, including multilayer perceptron (MLP), decision trees (DT), and Bayesian ridge regression (BRR), were applied. Also, the MLP is optimized to enhance the prediction by employing the Adam optimization to propose OMLP. The OMLP is compared to regular MLP, DT, and BRR using a variety of evaluation metrics, including statistical measures, the goodness of fit, and regression accuracy. The OMLP achieves better performance of dependent variables; Y HDH , Y SS , T swh,in , T swh,o , T a,H,in , T a,H,o , and Φ a,H,in using whole independent variables by RMSE ∈ [0.07171, 8.05636], Max Error ∈ [0.10108, 14.19858], and CRM ∈ [0.0202, 0.0329] and using top three correlated independent variables by MAE ∈ [0. 0514, 5.777], RMSE ∈ [0.07171, 7.0559], and CRM ∈ [0.0029, 0.0498]. [Display omitted] • Analysis experimental dataset and prepare for regression using artificial intelligence. • Design multiple single regression instead of a single multiple regression model. • Normalize and outlier pre-processing to overcome model training fitting problem. • Enhance each regression model based on Adam hyperparameters optimization. • Select top-K intra-correlated & inter-correlated variables by features subset evaluator. • Validate experimental insights using different performance metrics. [ABSTRACT FROM AUTHOR]

This paper is concerned with the use of regression methods to predict values of a response variable when that variable is naturally ordered. An application to the prediction of student examination performance is provided and it is argued that, although individual scores are unlikely to be well predicted at the extremes of the range using the conditional mean, conditional on covariates, it is possible to usefully predict where an individual is likely to feature in the rank order of performance. [ABSTRACT FROM AUTHOR]

The implementation of neural network regression prediction based on digital circuits is one of the challenging problems in the field of machine learning and cognitive recognition, and it is also an effective way to relieve the pressure of the Internet in the era of intelligence. As a nonlinear network, the stochastic configuration network (SCN) is considered to be an effective method for regression prediction due to its good performance in learning and generalization. Therefore, in this paper, we adapt the SCN to regression analysis, and design and verify the field programmable gate array (FPGA) framework to implement SCN model for the first time. In addition, in order to improve the performance of the SCN model based on the FPGA, the implementation of the nonlinear activation function on the FPGA is optimized, which effectively improves the prediction accuracy while considering the utilization rate of hardware resources. Experimental results based on the simulation data set and the real data set prove that the proposed FPGA framework successfully implements the SCN regression prediction model, and the improved SCN model has higher accuracy and a more stable performance. Compared with the extreme learning machine (ELM), the prediction performance of the proposed SCN implementation model based on the FPGA for the simulation data set and the real data set is improved by 56.37% and 17.35%, respectively. [ABSTRACT FROM AUTHOR]

Nanocellulose-modified presspaper is a promising solution to achieve cellulose insulation with better performance, reducing the risk of electrical insulation failures of a converter transformer. Predicting the dielectric properties will help to further design and improvement of presspaper. In this paper, a multivariable method was adopted to determine the effect of softwood fiber on the macroscopic performance of presspaper. Based on the parameters selected using the optimum subset method, a multiple linear regression was built to model the relationship between the fiber properties and insulating performance of presspaper. The results show that the fiber width and crystallinity had an obvious influence on the mechanical properties of presspaper, and fiber length, fines, lignin, and nanocellulose had a significant impact on the breakdown properties. The proposed models exhibit a prediction accuracy of higher than 90% when verified with the experimental results. Finally, the effect of nanocellulose on the breakdown strength of presspaper was taken into account and new models were derived.