Neural Network-based Prediction Research Articles

Temporal Convolutional Neural Network-Based Prediction of Vascular Health in Elderly Women Using Photoplethysmography-Derived Pulse Wave during Exercise.

(1) Background: The objective of this study was to predict the vascular health status of elderly women during exercise using pulse wave data and Temporal Convolutional Neural Networks (TCN); (2) Methods: A total of 492 healthy elderly women aged 60-75 years were recruited for the study. The study utilized a cross-sectional design. Vascular endothelial function was assessed non-invasively using Flow-Mediated Dilation (FMD). Pulse wave characteristics were quantified using photoplethysmography (PPG) sensors, and motion-induced noise in the PPG signals was mitigated through the application of a recursive least squares (RLS) adaptive filtering algorithm. A fixed-load cycling exercise protocol was employed. A TCN was constructed to classify flow-mediated dilation (FMD) into "optimal", "impaired", and "at risk" levels; (3) Results: TCN achieved an average accuracy of 79.3%, 84.8%, and 83.2% in predicting FMD at the "optimal", "impaired", and "at risk" levels, respectively. The results of the analysis of variance (ANOVA) comparison demonstrated that the accuracy of the TCN in predicting FMD at the impaired and at-risk levels was significantly higher than that of Long Short-Term Memory (LSTM) networks and Random Forest algorithms; (4) Conclusions: The use of pulse wave data during exercise combined with the TCN for predicting the vascular health status of elderly women demonstrated high accuracy, particularly in predicting impaired and at-risk FMD levels. This indicates that the integration of exercise pulse wave data with TCN can serve as an effective tool for the assessment and monitoring of the vascular health of elderly women.

Deep neural network-based prediction of tsunami wave attenuation by mangrove forests

The goal of this research is to develop a model employing deep neural networks (DNNs) to predict the effectiveness of mangrove forests in attenuating the impact of tsunami waves. The dataset for the DNN model is obtained by simulating tsunami wave attenuation using the Boussinesq model with a staggered grid approximation. The Boussinesq model for wave attenuation is validated using laboratory experiments exhibiting a mean absolute error (MAE) ranging from 0.003 to 0.01. We employ over 40,000 data points generated from the Boussinesq numerical simulations to train the DNN. Efforts are made to optimize hyperparameters and determine the neural network architecture to attain optimal performance during the training process. The prediction results of the DNN model exhibit a coefficient of determination (R2) of 0.99560, an MAE of 0.00118, a root mean squared error (RMSE) of 0.00151, and a mean absolute percentage error (MAPE) of 3%. When comparing the DNN model with three alternative machine learning models— support vector regression (SVR), multiple linear regression (MLR), and extreme gradient boosting (XGBoost)— the performance of DNN is superior to that of SVR and MLR, but it is similar to XGBoost.•High-accuracy DNN models require hyperparameter optimization and neural network architecture selection.•The error of DNN models in predicting the attenuation of tsunami waves by mangrove forests is less than 3%.•DNN can serve as an alternate predictive model to empirical formulas or classical numerical models.

Convolutional Neural Network-Based Prediction of Axial Length Using Color Fundus Photography.

To develop convolutional neural network (CNN)-based models for predicting the axial length (AL) using color fundus photography (CFP) and explore associated clinical and structural characteristics. This study enrolled 1105 fundus images from 467 participants with ALs ranging from 19.91 to 32.59 mm, obtained at National Taiwan University Hospital between 2020 and 2021. The AL measurements obtained from a scanning laser interferometer served as the gold standard. The accuracy of prediction was compared among CNN-based models with different inputs, including CFP, age, and/or sex. Heatmaps were interpreted by integrated gradients. Using age, sex, and CFP as input, the mean ± standard deviation absolute error (MAE) for AL prediction by the model was 0.771 ± 0.128 mm, outperforming models that used age and sex alone (1.263 ± 0.115 mm; P < 0.001) and CFP alone (0.831 ± 0.216 mm; P=0.016) by 39.0% and 7.31%, respectively. The removal of relatively poor-quality CFPs resulted in a slight MAE reduction to 0.759 ± 0.120 mm without statistical significance (P=0.24). The inclusion of age and CFP improved prediction accuracy by 5.59% (P=0.043), while adding sex had no significant improvement (P=0.41). The optic disc and temporal peripapillary area were highlighted as the focused areas on the heatmaps. Deep learning-based prediction of AL using CFP was fairly accurate and enhanced by age inclusion. The optic disc and temporal peripapillary area may contain crucial structural information for AL prediction in CFP. This study might aid AL assessments and the understanding of the morphologic characteristics of the fundus related to AL.

Backpropagation Neural Network-Based Prediction of Kovats Retention Index for Essential Oil Compounds

Backpropagation Neural Network-Based Prediction of Kovats Retention Index for Essential Oil Compounds

Performance of the neural network-based prediction model in closed-loop adaptive optics.

Adaptive optics (AO) technology is an effective means to compensate for atmospheric turbulence, but the inherent delay error of an AO system will cause the compensation phase of the deformable mirror (DM) to lag behind the actual distortion, which limits the correction performance of the AO technology. Therefore, the feed-forward prediction of atmospheric turbulence has important research value and application significance to offset the inherent time delay and improve the correction bandwidth of the AO system. However, most prediction algorithms are limited to an open-loop system, and the deployment and the application in the actual AO system are rarely reported, so its correction performance improvement has not been verified in practice. We report, to our knowledge, the first successful test of a deep learning-based spatiotemporal prediction model in an actual 3 km laser atmospheric transport AO system and compare it with the traditional closed-loop control methods, demonstrating that the AO system with the prediction model has higher correction performance.

A recompression supercritical CO2 cycle using heat source storage to achieve low-frequency reactor regulation: Analysis and optimization of load capacity under safe compressor operation

This study focuses on proposing a heat source storage mode for the indirect recompression supercritical CO2 cycle (RSCC) coupled with a sodium-cooled fast reactor, aiming to achieve low-frequency regulation of the reactor and enhance the load capacity of system. A deep neural network-based turbomachinery performance prediction approach is employed to assess the system's part-load and overload performance under several typical control strategies. The turbine inlet temperature of the system under all control strategies except the intermediate heat exchanger (IHX) bypass rises with decreasing load. The system's part-load performance is positively correlated with its corresponding overload power. The findings reveal that the adjustable load ranges of the heat source storage tanks under inventory, turbine bypass, turbine throttle, and IHX bypass controls are 10%–156.10%, 10%–156.10%, 30%–147.30%, and 10%–146.88%, respectively, along with the safe compressor operation. As the near-choke operation of the compressor limits the rise in overload power, a turbine fitted with variable inlet guide vanes is tried. The optimization results demonstrate that the largest overload power of the system can reach 161.18 % without reducing the thermal efficiency. This study provides a heuristic insight into the variable load operation for indirect nuclear-powered Brayton cycles under low-frequency reactor regulation.

Traffic noise measurement, mapping, and modeling using soft computing techniques for mid-sized smart Indian city

The present study presents an investigation into the utilization of artificial neural networks (ANN) and multiple linear regression model (MLR) for the prediction of traffic noise levels in various locations of Dhanbad city at varying intervals. Traffic noise indices measurements were carried out using sound pressure level meter (SLM). The prediction of equivalent A-weighted sound level (LAeq) and the sound level exceeding 10 percent of the time (L10) is carried out using various influencing factors such as number of cars, number of 2 wheelers, number of 3 wheelers, number of heavy vehicle, number of medium commercial vehicles, and traffic speed. The findings demonstrate the ANN model's proficiency in comparison of MLR model for providing precise prediction accuracy of traffic noise indices with a R2 of 0.94 for LAeq and 0.91 for L10. Furthermore, the frequency spectrum analysis reveals that high peaks were observed at lower frequencies ranging from 31.5 Hz to 50 Hz, middle frequencies from 500 to 800 Hz and higher frequencies from 3.5 kHz to 5 kHz. The noise maps at varying intervals revealed that most of the locations are having higher noise levels due to increase in vehicular movement. The study emphasizes on the potential significance of the proposed neural network-based prediction model in collaboration with noise mapping as a vital tool for the anticipation of traffic noise levels and the formulation of noise mitigation strategies in context of the mid-sized smart city like Dhanbad, India.

Convolutional neural network-based wind pressure prediction on low-rise buildings

A new model of wind pressure prediction for low-rise buildings is built based on deep convolutional neural network (CNN), and the model is trained and tested by the aerodynamic data from international open database. The prediction accuracy is compared with that of the literature, and then comprehensive error analysis on all 335 predicted taps under 3 untrained wind directions are carried out for mean and RMS pressure coefficients by various error metrics. The study shows that the distributions of mean and RMS coefficients predicted by the CNN model are closer to the experimental results in comparison with those predicted by the traditional artificial neural network (ANN) model and the common deep neural network (DNN) model without the convolution and pooling layers. The relative error on the corner tap is 13.5%, 4.3% and 0.1% by the ANN, DNN, and CNN models respectively. The mean-square-error (MSE) of the corner bay and the whole roof are also eminently reduced by the CNN model. By introducing the convolutional and pooling layers, more prediction errors are confined within 5% and basically no errors exceed 20%. Consequently, the CNN algorithm can be applied in solving various wind pressure prediction problems on buildings in the future work.

Artificial Neural Network-Based Prediction of Physical and Mechanical Properties of Concrete Containing Glass Aggregates

This comprehensive study analyzes the use of crushed glass as both fine and coarse aggregate in concrete, as well as the prediction accuracy of Artificial Neural Networks (ANN). The primary objectives are to understand the interactions between concrete’s constituents and to assess the accuracy of ANN models in predicting concrete’s mechanical and physical properties. This is achieved using a two-decade experimental results dataset of concrete’s compressive and tensile strengths, slump, density, and the corresponding mix design proportions, including waste glass aggregate. A series of 70 concrete samples were carefully built and tested, with compressive strengths varying from 12 to 71 MPa and glass aggregate percentages ranging from 0-100%. These samples served as the basis for the creation of an input dataset and ANN targets. The ANN model underwent intensive training, validation, testing, and statistical regression analysis. The ANN models are exceptionally accurate, with a continuously low error margin of roughly 2%, highlighting their usefulness in matching experimental and predicted results. Validation techniques highlight the models' dependability, with consistently high coefficients of determination (R-values), including 0.99484, demonstrating their robustness in replicating complicated concrete properties. The data analysis shows a unique pattern, with optimum glass aggregate percentages in the range of 10–20%. Beyond this range, there is a noticeable decline in concrete properties. Finally, the study confirms the efficacy of ANN in predictive modeling while also validating the potential of crushed glass to replace natural aggregates in concrete. Doi: 10.28991/CEJ-2024-010-05-018 Full Text: PDF

Neural Network-Based Body Weight Prediction in Pelibuey Sheep through Biometric Measurements

This paper presents an intelligent system for the dynamic estimation of sheep body weight (BW). The methodology used to estimate body weight is based on measuring seven biometric parameters: height at withers, rump height, body length, body diagonal length, total body length, semicircumference of the abdomen, and semicircumference of the girth. A biometric parameter acquisition system was developed using a Kinect as a sensor. The results were contrasted with measurements obtained manually with a flexometer. The comparison gives an average root mean square error (RMSE) of 9.91 and a mean R2 of 0.81. Subsequently, the parameters were used as input in a back-propagation artificial neural network. Performance tests were performed with different combinations to make the best choice of architecture. In this way, an intelligent body weight estimation system was obtained from biometric parameters, with a 5.8% RMSE in the weight estimations for the best architecture. This approach represents an innovative, feasible, and economical alternative to contribute to decision-making in livestock production systems.

Parametric study and neural network-based prediction for stress concentration factor of concrete-filled steel tubular T-joint

In this study, an extensive parametric investigation for the stress concentration factor (SCF) of concrete-filled steel tubular (CFST) T-joints was conducted based on a validated finite element (FE) model in which the weld profile, mesh, and contact simulation were specifically and precisely dealt with. Distribution of the SCF on the intersecting line was depicted. Effect of non-dimensional geometric parameters and the elastic modulus of concrete was discussed in detail. Results show that the SCF is monotonically distributed on the intersecting line in the chord side, and the maximum SCF probably located at the crown position. There is the possibility that the maximum SCF of the brace occurs at a position neither the crown nor the saddle. Reducing the length of chord or promoting the elastic modulus of concrete lowers the SCF at most positions, but has no effect on the SCF at chord saddle. A back-propagation neural network (BPNN) model was established, and was trained based on 724 FE results of the SCF from the parametric study. The trained BPNN model predicts the SCF of CFST T-joints with a low error of 14.5%, and it improves the accuracy by 62.0% compared to the current formulae.

Neural Network-Based Prediction of Perceived Sleep Quality Through Wearable Device Data.

This study focuses on the development of a neural network model to predict perceived sleep quality using data from wearable devices. We collected various physiological metrics from 18 participants over four weeks, including heart rate, physical activity, and both device-measured and self-reported sleep quality. The primary objective was to correlate wearable device data with subjective sleep quality perceptions. Our approach used data processing, feature engineering, and optimizing a Multi-Layer Perceptron classifier. Despite comprehensive data analysis and model experimentation, the predictive accuracy for perceived sleep quality was moderate (59%), highlighting the complexities in accurately quantifying subjective sleep experiences through wearable data. Applying a tolerance of 1 grade (on a scale from 1-5), increased accuracy to 92%. More in-depth analysis is required to fully comprehend how wearables and artificial intelligence might assist in understanding sleep behavior.

Cross-project concurrency bug prediction using domain-adversarial neural network

In recent years, software bug prediction has shown to be effective in narrowing down the potential bug modules and boosting the efficiency and precision of existing testing and analysis tools. However, due to its non-deterministic nature and low presence, concurrency bug labeling is a challenging task, which limits the implementation of within-project concurrency bug prediction. This paper proposes DACon, a Domain-Adversarial neural network-based cross-project Concurrency bug prediction approach to tackle this problem by leveraging information from another related project. By combining a set of designed concurrency code metrics with widely used sequential code metrics, DACon uses SMOTE (Synthetic Minority Over-sampling TEchnique) and domain-adversarial neural network to mitigate two challenges including the severe class imbalance between concurrency bug-prone samples and concurrency bug-free samples, and shift between source and target distribution during bug prediction implementation. Our evaluation on 20 pair-wise groups of experiments constructed from 5 real-world projects indicates that cross-project concurrency bug prediction is feasible, and DACon can effectively predict concurrency bugs across different projects.

An Overview of Demand Analysis and Forecasting Algorithms for the Flow of Checked Baggage among Departing Passengers

The research on baggage flow plays a pivotal role in achieving the efficient and intelligent allocation and scheduling of airport service resources, as well as serving as a fundamental element in determining the design, development, and process optimization of airport baggage handling systems. This paper examines baggage checked in by departing passengers at airports. The crrent state of the research on baggage flow demand is first reviewed and analyzed. Then, using examples of objective data, it is concluded that while there is a significant correlation between airport passenger flow and baggage flow, an increase in passenger flow does not necessarily result in a proportional increase in baggage flow. According to the existing research results on the influencing factors of baggage flow sorting and classification, the main influencing factors of baggage flow are divided into two categories: macro-influencing factors and micro-influencing factors. When studying the relationship between the economy and baggage flow, it is recommended to use a comprehensive analysis that includes multiple economic indicators, rather than relying solely on GDP. This paper provides a brief overview of prevalent transportation flow prediction methods, categorizing algorithmic models into three groups: based on mathematical and statistical models, intelligent algorithmic-based models, and combined algorithmic models utilizing artificial neural networks. The structures, strengths, and weaknesses of various transportation flow prediction algorithms are analyzed, as well as their application scenarios. The potential advantages of using artificial neural network-based combined prediction models for baggage flow forecasting are explained. It concludes with an outlook on research regarding the demand for baggage flow. This review may provide further research assistance to scholars in airport management and baggage handling system development.

Effect of Conformational Variability on the Drug Resistance of Candida auris ERG11p and FKS1.

Candida auris infection has been recognized as an urgent threat to antifungal drug resistance, and the Eagle effect of C. auris FKS1 (1,3-β-d-glucan synthase) wild-type isolates has also been noted. The Eagle effect, namely, where higher concentrations of antifungals reduce fungicidal activity relative to lower concentrations, is a confounding factor of apparent antifungal resistance, but the detailed mechanism remains unclear. Here, we present the conformational variability of mutation sites for ERG11p (lanosterol 14α-demethylase) and FKS1 from deep neural network-based prediction along with the reported X-ray crystallographic and cryo-electron microscopy (cryo-EM) structures of antifungals. The sequence variability maps provide valuable insights into the inconsistent correlation between azole resistance and the mysterious Eagle effect with the dispersion of minimal inhibitory concentration (MIC) for echinocandin resistance. The conformational variability prediction supports the hypothesis that mutations K143R of clade I, VF125AL of clade III, and Y132F of clade IV for C. auris ERG11p make the corresponding site variable and that an increased population of invisible variable conformations potentially contributes to triazole resistance. In contrast, the predicted rigid conformation by the S639F mutation of hot spot region 1 (HS1) for FKS1 suggests that caspofungin (CAS) is involved in an uncompetitive inhibition, and a decreased population of the CAS-bound state of FKS1 with Rho1 leads to drug resistance. The predicted variable HS1 region for FKS1 WT isolates and the rigid one for FKS1 S639F mutants support the in vivo drug response and the in vitro MIC dispersion. A plausible mechanism of the Eagle effect is hereby proposed, namely, that a high concentration of CAS with a high membrane affinity reduces the population of the CAS-bound state of FKS1 with Rho1, as well as accompanying events such as aggregation or association depending on the conformational variability of HS1.

Enhancing office building energy efficiency: neural network-based prediction of energy consumption

Enhancing office building energy efficiency: neural network-based prediction of energy consumption

Neural Network-Based prediction for Cross-Temperature induced VT distribution shift in 3D NAND flash memory

In this study, a neural network (NN) was proposed for predicting the VT characteristics of NAND flash memories under cross-temperature conditions. The training data were obtained from commercial NAND flash memory chip measurements at various temperatures. The VT distribution shift caused by cross-temperature was accurately predicted by investigating the optimum data dimensions while minimizing the data generation process. Two types of NNs were used to achieve an accurate VT distribution prediction, and each network was optimized using specific parameters based on the data characteristics at various program verify levels. Finally, quantitative and visual evaluations were conducted to verify the performance of the trained NNs. When the program-measured temperature varied from low to high, the NNs achieved mean errors of 1.87%, 1.41% at low and 0.34%, 0.77% at high for the average and width of the VT distribution, respectively. Similarly, when the temperature varied from high to low, the corresponding mean errors were 2.01%, 0.74% at high and 0.23%, 1.59% at low. These findings demonstrate that NNs can minimize the procedures for detecting the VT distribution shift caused by cross-temperature, thereby offering a promising approach to enhance reliability in the presence of such effects.

Determining the minimum data size for the development of artificial neural network-based prediction models for rice pests in Korea

Sudden outbreaks of crop pests (insect pests and diseases) are increasing in Korea due to climate change and globalization. To prevent such outbreaks, it is necessary to predict and control pest occurrences in advance. Crop pests have been predicted through process-based or statistical modeling; however, the limitations of these models, which rely solely on historically acquired domain knowledge and data, have become increasingly prominent owing to climate change and rapidly changing agricultural ecosystems. To overcome these limitations, artificial neural network (ANN)-based models that use continuous pest survey data in the field have been investigated over the last decade. However, because pest survey data are collected by humans through process-mediated methods, fundamental problems exist in terms of data quality and size that may hinder the performance of the resulting ANN-based models. In this study, to determine the minimum pest data size required to ensure the optimal performance of ANN-based models, we developed feed-forward neural network models for 19 rice pests using 23 pest datasets collected from 149 districts by the Rural Development Administration of Korea over 19 years (2002–2020). Using each ANN-based model, the minimum data size required for the highest model performance achieved in this study was determined for all 19 rice pests. Furthermore, we developed a decision-tree rule to estimate the minimum data size based on the selected characteristics of each pest. The final Decision tree rule, based on the distinction between diseases and insect pests and the balance of pest data (the relative ratio of pest occurrence data to non-occurrence data), showed a relatively good performance (70.24 %) in the 3-fold cross-validation test. Overall, these results indicate that the minimum data sizes required for ANN modeling vary among rice pests, depending on the pest data characteristics, as indicated by the Decision tree rule developed in this study.

Neural network-based prediction of auto-ignition temperature of ternary mixed liquids

Auto-ignition temperature (AIT) is one of the crucial exponents in the design of fire and explosion safety measures. Therefore, in this study, quantitative structure-property relationship approach was used to predict the AIT of ternary hybrid liquids based on molecular structure information. The optimal molecular descriptors were calculated and filtered using Mordred software. Twelve mixing rules were proposed for calculating molecular descriptors of mixtures. A prediction model for the AIT value of binary liquid mixtures was developed, validated and evaluated using a back propagation neural network (BPNN) and a one-dimensional convolutional neural network (1DCNN). The relative contribution and positive and negative correlations between individual molecular descriptors and AIT in the model were interpreted using the shapley additive explanations method. The results show that BPNN and 1DCNN models using mixing rule 1 have the best fitting ability, stability and prediction ability. The determination coefficient of the BPNN and 1DCNN models in the training set were 0.996 and 0.992, the root mean square errors were 3.613 °C and 5.284 °C, the mean absolute errors were 2.483 °C and 4.144 °C, the nash efficiency coefficient was 0.996 and 0.992, respectively, the willmott index was 0.999 and 0.998. and the values of the top three molecular descriptors of relative contribution, SssCH2, SsOH and SsCH3, were negatively correlated with the AIT values. The BPNN and 1DCNN models provide an accurate and reliable method for predicting ternary mixing liquid AIT.

Zero-Shot Task Adaptation with Relevant Feature Information

We propose a method to learn prediction models such as classifiers for unseen target tasks where labeled and unlabeled data are absent but a few relevant input features for solving the tasks are given. Although machine learning requires data for training, data are often difficult to collect in practice. On the other hand, for many applications, a few relevant features would be more easily obtained. Although zero-shot learning or zero-shot domain adaptation use external knowledge to adapt to unseen classes or tasks without data, relevant features have not been used in existing studies. The proposed method improves the generalization performance on the target tasks, where there are no data but a few relevant features are given, by meta-learning from labeled data in related tasks. In the meta-learning phase, it is essential to simulate test phases on target tasks where prediction model learning is required without data. To this end, our neural network-based prediction model is meta-learned such that it correctly responds to perturbations of the relevant features on randomly generated synthetic data. By this modeling, the prediction model can explicitly learn the discriminability of the relevant features without real target data. When unlabeled training data are available in the target tasks, the proposed method can incorporate such data to boost the performance in a unified framework. Our experiments demonstrate that the proposed method outperforms various existing methods with four real-world datasets.