Neural Network Algorithm Research Articles

Full-field expansion and damage detection from sparse measurements using physics-informed variational autoencoders

Recent advancements in machine learning and artificial neural network algorithms have created new opportunities for expanding sparse measurements to full-field representations. However, these methods often face accuracy challenges when used for condition monitoring of physical systems. For example, traditional autoencoders can capture complex phenomena and their underlying physics in the latent space, but due to their frequentist nature, they lack generative and damage detection capabilities. To address these challenges, a novel physics-informed variational autoencoder (PI-VAE) network is proposed for expanding sparse measurements to full-field representations while also detecting damage. The effectiveness of the proposed PI-VAE network is evaluated through analytical and experimental studies on a metal plate under thermal excitation with embedded defects of various sizes and types. In the analytical studies using finite-element model data, the PI-VAE accurately expanded full-field temperature distributions and identified the dimensions of cracks, spalling, and hole-like defects with errors smaller than 5%. When tested with experimental data, the PI-VAE network maintained robust performance, detecting damage with errors smaller than 6%, despite being trained on undamaged data only. These findings demonstrate the PI-VAE’s potential as a reliable tool for full-field expansion and damage detection in structural health monitoring and nondestructive evaluation, even when limited sensors and datasets are available.

Comparison of machine learning algorithms for dengue virus (DENV) classification

Dengue, transmitted by the Aedes aegypti mosquito, necessitates accurate case classification for effective management. The research focuses on leveraging machine learning algorithms to enhance diagnosis and streamline control strategies. Utilizing a dataset from a public hospital, encompassing 21,157 cases categorized by period, outcome, gender, age, symptoms, and origin (autochthonous or imported), the study conducted a comparative analysis of Support Vector Machine, Random Forest, and Artificial Neural Network algorithms. The dataset was divided into 70% (14,809 cases) for training and 30% (6,348 cases) for testing. Results unveiled the Artificial Neural Network as the frontrunner, exhibiting an impressive 86.47% accuracy and a robust 92.91% recall in the classification of dengue-related cases. This underscores the potential of machine learning in refining dengue diagnosis and facilitating more efficient control strategies, offering a promising avenue for combating this vector-borne disease.

Prediction of Spatiotemporal Pollution of Soil Heavy Metal in Mining Areas Based on Grey Neural Network Algorithm

Prediction of Spatiotemporal Pollution of Soil Heavy Metal in Mining Areas Based on Grey Neural Network Algorithm

Machine learning‐driven microwave imaging for soil moisture estimation near leaky pipe

AbstractCharacterizing soil moisture around drip irrigation pipes is crucial for precise and optimized farming. Machine learning (ML) approaches are particularly suitable for this task as they can reduce uncertainties caused by soil conditions and the drip pipe positions, using features extracted from relevant datasets. This letter addresses local moisture detection in the vicinity of dripping pipes using a portable microwave imaging system. The employed ML approach is fed with two dimensional images generated using back projection as a radar‐based algorithm and the Born approximation as an inverse scattering method, based on spatio‐temporal (collected data at various positions over the soil surface and at different time points.) measurements at various frequencies. The study investigates the performance of K‐nearest neighbour (KNN) and convolutional neural networks (CNN) algorithms for moisture classification based on these imaging techniques. We also explore the potential of KNN and CNN for moisture estimation around the plant roots and in the presence of pebbles. In general, CNN outperforms KNN in moisture content detection from microwave data, especially after applying imaging algorithms. A combination of CNN as the ML approach and the back projection algorithm to provide training data, yielded accuracy more than other models for moisture content estimation. Also, the practical results demonstrate that our method can detect soil moisture with an estimation error of less than 10%.

Surface Mechanical Property Prediction and Process Optimization of 18CrNiMo7-6 Carburized Steel Stator Guide Based on Radial Basis Function Neural Network and NSGA-II Algorithm

The carburizing process is a key technology that affects the mechanical properties of the surface of the hydraulic motor stator guide rail, and the related process parameters have an important influence on surface hardness, the thickness of the carburized layer, and the deformation of the guide rail. However, at present, the relationship between the carburizing process parameters and the surface mechanical properties of the target is not clear. This paper proposes a “hardness prediction and process parameter optimization” method. Firstly, a finite element model is established, with carburizing time, temperature, and carbon potential as the three input factors; the optimal Latin hypercubic experimental design and sensitivity analysis are applied. Secondly, surface hardness, carburized layer thickness, and deformation are taken as the output values, and an RBF neural network is used to construct the prediction model. The results show that the RBF neural network can be accurately used for the prediction of surface hardness, the thickness of the carburized layer, and deformation, and for the optimization of process parameters. The optimized parameters of surface hardness and the thickness of the carburized layer were increased by 4.2% and 5.1%, respectively, and the deformation amount was reduced to 0.31 mm, achieving the goal of optimal design.

Prediction of the Infrared Absorbance Intensities and Frequencies of Hydrocarbons: A Message Passing Neural Network Approach.

Accurately and efficiently predicting the infrared (IR) spectra of a molecule can provide insights into the structure-properties relationships of molecular species, which has led to a proliferation of machine learning tools designed for this purpose. However, earlier studies have focused primarily on obtaining normalized IR spectra, which limits their potential for a comprehensive analysis of molecular behavior in the IR range. For instance, to fully understand and predict the optical properties, such as the transparency characteristics, it is necessary to predict the molar absorptivity IR spectra instead. Here, we propose a graph-based communicative message passing neural network algorithm that can predict both the peak positions and absolute intensities corresponding to density functional theory calculated molar absorptivities in the IR domain. By modifying existing spectral loss functions, we show that our method is able to predict with DFT-accuracy level the IR molar absorptivities of a series of hydrocarbons containing up to ten carbon atoms and apply the model to a set of larger molecules. We also compare the predicted spectra with those generated by the direct message passing neural network. The results suggest that both algorithms demonstrate similar predictive capabilities for hydrocarbons, indicating that either model could be effectively used in future research on spectral prediction for such systems.

Characterizing the progression of subclinical cardiac amyloidosis through artificial intelligence applied to electrocardiographic images and echocardiograms

Abstract Background Transthyretin amyloid cardiomyopathy (ATTR-CM) follows an indolent course and may remain undiagnosed for many years before resulting in clinical disease. Purpose To assess the utility of scalable artificial intelligence (AI) methods applied to electrocardiographic (ECG) images and echocardiograms in detecting longitudinal changes that precede the clinical diagnosis of ATTR-CM. Methods Across 5 hospitals of a large U.S.-based hospital system, we identified a total of 234 patients (age 79 [IQR:70-85] years, n=56 (23.9%) women) with ATTR-CM, defined by the presence of a positive bone scintigraphy scan (or pharmacotherapy with an approved transthyretin stabilizer) between 2015 and 2023. Using 10:1 age- and sex-matched controls, we independently trained convolutional neural network algorithms to discriminate ATTR-CM from controls using i) ECG images (AUROC 0.915) or ii) echocardiographic studies by incorporating all available views (AUROC 0.93). In a leave-out testing set of n=82 individuals with serial ECGs and n=49 individuals with serial echocardiograms we characterized longitudinal patterns in the evolution of the AI ECG and echo-based scores, and estimated the percentage of individuals that would have screened positive across distinct time intervals. Results In the leave-out testing set 52.4% and 34.7% of all individuals who went on to receive a diagnosis of ATTR-CM would have screened positive by AI deployed to ECGs and echocardiograms, respectively, 1 to 3 years before their confirmatory testing, with the percentage increasing to 75.6% and 69.4% in the 12 months before their eventual diagnosis (A). Across both modalities, there was a longitudinal increase in AI predictions when comparing the median of all studies performed more than a year before diagnosis versus the median of any studies performed within 12 months of diagnosis (AI-Echo: 0.20 [IQR: 0.04-0.35] to 0.49 [0.24-0.59] & AI-ECG: 0.07 [0.02-0.20] to 0.32 [0.11-0.49], p<0.001 for both) (B), consistent with evolution in the ECG and echocardiographic signature of ATTR-CM before its clinical diagnosis. Conclusions We demonstrate that AI applied directly to ECG images and echocardiography may enable scalable monitoring of subclinical ATTR-CM progression.

A Model for Fat Content Detection in Walnuts Based on Near-Infrared Spectroscopy

Near-infrared spectroscopy (NIR) is an efficient and accurate method for fat content detection in walnuts. ‘Wen 185’ walnut is grown in large quantities in southern Xinjiang, and its fat content is an important indicator for evaluating the intrinsic quality. The excessive pursuit of yield efficiency, combined with the neglect of quality, agricultural product safety and other factors, has led to the production of poor-quality walnuts. Moreover, research on predicting walnut kernel fat content based on near-infrared spectroscopy technology is rarely reported. Therefore, a technical framework for walnut kernel detection based on near-infrared spectroscopy and the technical standards for ‘Wen 185’ are urgently needed. After first optimizing the initial spectrum data using five preprocessing methods, we established separate prediction models for walnut kernel fat content based on either a back propagation neural network or a support vector regression (SVR) algorithm. The results show that the correction set and validation set coefficients of determination of the walnut kernel fat content prediction model using the back propagation neural network algorithm were 0.86 and 0.89, respectively, with root mean square errors of 1.56 and 1.58, and an RPD value of 2.57; the coefficients of determination for the calibration and validation sets of SVR were 0.90 and 0.83, respectively, with root mean square errors of 1.76 and 1.70, respectively, and an RPD value of 1.70. Thus, near-infrared spectroscopy can be used to establish a foundation for realizing the rapid detection of walnut kernel fat content.

Multi-objective optimization of switched reluctance motor driving electric vehicles based on vehicle operating condition

To improve the comprehensive performance of the switched reluctance motor (SRM) for electric vehicles (EVs) under actual driving conditions, a novel geometrical optimization method based on the time-varying phase current characteristics and the comprehensive indicator with justified weight coefficients is proposed in this paper. First, based on the established dynamic models of EV and SRM, the actual operating phase currents of the motor working under the New European Driving Cycle (NEDC) are extracted and classified to develop the representative optimization conditions. Then, considering the requirements of EV, five indicators are defined to evaluate the comprehensive performance of SRM, and a new multi-objective synchronization optimization function is proposed with five weight coefficients obtained by the statistical analysis method for the sample data of the indicators. Furthermore, the geometrical optimization for a four-phase 8/6 SRM is carried out by the combination of Particle Swarm Optimization (PSO) and BP neural network algorithms, after analyzing the effects of key geometrical parameters on the indicators. Finally, the work performance of the optimized SRM is evaluated and compared with that of the initial prototype, and the results show that the optimized SRM with the proposed multi-objective synchronization optimization method can improve the comprehensive indicator by 59.59%, and can significantly enhance the vehicle actual working economy by 50.74% under NEDC cycle.

An Improved Neural Network Algorithm for Energy Consumption Forecasting

Accurate and efficient forecasting of energy consumption is a crucial prerequisite for effective energy planning and policymaking. The BP neural network has been widely used in forecasting, machine learning, and various other fields due to its nonlinear fitting ability. In order to improve the prediction accuracy of the BP neural network, this paper introduces the concept of forecast lead time and establishes a mathematical model accordingly. Prior to training the neural network, the input layer data are preprocessed based on the forecast lead time model. The training and forecasting results of the BP neural network when and when not considering forecast lead time are compared and verified. The findings demonstrate that the forecast lead time model can significantly improve the prediction speed and accuracy, proving to be highly applicable for short-term energy consumption forecasting.

Spatial analysis and soft computational modeling for hazard assessment of potential toxic elements in potable groundwater

Swiftly increasing population and industrial developments of urban areas has accelerated the worsening of the water quality in recent years. Groundwater samples from different locations of the Doon valley, Garhwal Himalaya were analyzed to measure concentrations of six potential toxic elements (PTEs) viz. chromium (Cr), nickel (Ni), arsenic (As), molybdenum (Mo), cadmium (Cd), and lead (Pb) using Inductively Coupled Plasma Mass Spectrometer (ICP-MS) with the aim to study the spatial distribution and associated hazards. In addition, machine learning algorithms have been used for prediction of water quality and identification of influencing PTEs. The results inferred that the mean values (in the units of µg L−1) of analyzed PTEs were observed in the order of Mo (1.066) > Ni (0.744) > Pb (0.337) > As (0.186) > Cr (0.180) > Cd (0.026). The levels and computed risks of PTEs were found below the safe limits. The radial basis function neural network (RBF-NN) algorithms showed high level of accuracy in the predictions of heavy metal pollution index (HPI), heavy metal evaluation index (HEI), non-carcinogenic (N-CR) and carcinogenic (CR) parameters with determination coefficient values ranged from 0.912 to 0.976. However, the modified heavy metal pollution index (m-HPI) and contamination index (CI) predictions showed comparatively lower coefficient values as 0.753 and 0.657, respectively. The multilayer perceptron neural network (MLP-NN) demonstrated fluctuation in precision with determination coefficient between 0.167 and 0.954 for the prediction of computed indices (HPI, HEI, CI, m-HPI). In contrast, the proficiency in forecasting of non-carcinogenic and carcinogenic hazards for both sub-groups showcased coefficient values ranged from 0.887 to 0.995. As compared to each other, the radial basis function (RBF) model indicated closer alignments between predicted and actual values for pollution indices, while multilayer perceptron (MLP) model portrayed greater precision in prediction of health risk indices.

“CLADE-FINDER”: Candida auris Lineage Analysis Determination by Fourier Transform Infrared Spectroscopy and Artificial Neural Networks

In 2019, Candida auris became the first fungal pathogen included in the list of the urgent antimicrobial threats by the Centers for Disease Control (CDC). Short tandem repeat (STR) analysis and whole-genome sequencing (WGS) are considered the gold standard, and can be complemented by other molecular methods, for the genomic surveillance and clade classification of this multidrug-resistant yeast. However, these methods can be expensive and require time and expertise that are not always available. The long turnaround time is especially not compatible with the speed needed to manage clonal transmission in healthcare settings. Fourier transform infrared (FTIR) spectroscopy, a biochemical fingerprint approach, has been applied in this study to a set of 74 C. auris isolates belonging to the five clades of C. auris (I-V) in combination with an artificial neural network (ANN) algorithm to create and validate “CLADE-FINDER”, a tool for C. auris clade determination. The CLADE-FINDER classifier allowed us to discriminate the four primary C. auris clades (I-IV) with a correct classification for 96% of the samples in the validation set. This newly developed genotyping scheme can be reasonably applied for the effective epidemiological monitoring and management of C. auris cases in real time.

A real-world pharmacovigilance study of FDA Adverse Event Reporting System (FAERS) for gemcitabine

ABSTRACT Background Gemcitabine is widely used in the treatment of various cancers. This study aims to evaluate gemcitabine-associated adverse events (AEs) using the Food and Drug Administration Adverse Event Reporting System (FAERS) database. Research design and methods We analyzed data spanning from January 2004 to June 2023. Employing reporting odds ratio (ROR) and Bayesian confidence propagation neural network (BCPNN) algorithms, we identified AEs with positive signals in patients administered gemcitabine. Results Out of 16,623,939 reports, 23,645 involved gemcitabine as the ‘primary suspected (PS)’ resulting in 74,306 AEs. Consistent with the reports in the specification and clinical trials, thrombocytopenia, pyrexia, neutropenia, and anemia were the most common AEs. Notably, our study identified some unexpected AEs such as abdominal pain, pleural effusion, ascites, and gastrointestinal hemorrhage, among others. The most significant SOC was ‘Blood and lymphatic system disorders’. The median onset time for gemcitabine-related AEs was 24 days (interquartile range [IQR] 6–82 days), with most cases occurring within the initial 30 days following gemcitabine administration. Conclusion Gemcitabine is associated with a broad spectrum of AEs affecting multiple organ systems, with a notable incidence of hospitalization. The study highlights both expected and unexpected AEs, which could enhance future clinical applications and safety of gemcitabine.

Optimal approximations for the free boundary problems of the space-time fractional Black-Scholes equations using a combined physics-informed neural network

The combined physics-informed neural network is employed to deal with the free boundary problems of fractional Black-Scholes equations. The solution assumption and the loss function are determined, the transfer learning is borrowed, the combined neural network with data enhancement layer is designed, then the classical Black-Scholes model is numerically solved and the comparative analysis of numerical results under different neural networks is made. For further insight into the long-term memory of fluctuation, the free boundary problems of the space-time Black-Scholes equations under Caputo, Caputo-Fabrizio and Atangana-Baleanu-Caputo fractional derivatives are studied. The corresponding empirical analyses are presented and the optimal exercise boundaries of American put option are simulated. The market analysis shows that introducing fractional calculus tools and neural network algorithms into American put option pricing can yield more realistic prediction results. The work provides a viable method for subsequent researchers to study American option pricing using fractional calculus and neural networks combined with true market data and to deal with the free boundary problems in other research fields.

Modeling rate of penetration using hybridization of Artificial Neural Network and Artificial Fish Swarm algorithm

This study aims to develop a robust and efficient approach for predicting the Rate of Penetration (ROP) by integrating the Artificial Fish Swarm Algorithm (AFSA) with the Back Propagation (BP) algorithm in an Artificial Neural Network (ANN). The integration of AFSA with Back Propagation is the primary novelty of this work, as it seeks to achieve optimal weight and bias values for the ANN, thereby enhancing prediction accuracy. A dataset of 1944 data points from a vertical well in southern Algeria was utilized. The performance of the developed models was evaluated using correlation coefficient and root mean square error (RMSE). The results demonstrated that the AFSA-ANN model significantly outperforms conventional ANN, Support Vector Regression (SVR), Random Forest regressor (RFR), and Bourgoyne and Young’s model in terms of accuracy. Additionally, the application of leverage method revealed that only 0.98% of the data falls outside the model’s applicability domain, indicating that the data and the model are statistically valid and reliable. The findings of this study have significant implications for the oil and gas field, providing a more accurate and reliable method for predicting ROP, which can lead to more efficient drilling operations and reduced costs.

Terahertz Non-Destructive Testing of Porosity in Multi-Layer Thermal Barrier Coatings Based on Small-Sample Data

Accurately characterizing the internal porosity rate of thermal barrier coatings (TBCs) was essential for prolonging their service life. This work concentrated on atmospheric plasma spray (APS)-prepared TBCs and proposed the utilization of terahertz non-destructive detection technology to evaluate their internal porosity rate. The internal porosity rates were ascertained through a metallographic analysis and scanning electron microscopy (SEM), followed by the reconstruction of the TBC model using a four-parameter method. Terahertz time-domain simulation data corresponding to various porosity rates were generated employing the time-domain finite difference method. In simulating actual test signals, white noise with a signal-to-noise ratio of 10 dB was introduced, and various wavelet transforms were utilized for denoising purposes. The effectiveness of different signal processing techniques in mitigating noise was compared to extract key features associated with porosity. To address dimensionality challenges and further enhance model performance, kernel principal component analysis (kPCA) was employed for data processing. To tackle issues related to limited sample sizes, this work proposed to use the Siamese neural network (SNN) and generative adversarial network (GAN) algorithms to solve this challenge in order to improve the generalization ability and detection accuracy of the model. The efficacy of the constructed model was assessed using multiple evaluation metrics; the results indicate that the novel hybrid WT-kPCA-GAN model achieves a prediction accuracy exceeding 0.9 while demonstrating lower error rates and superior predictive performance overall. Ultimately, this work presented an innovative, convenient, non-destructive online approach that was safe and highly precise for measuring the porosity rate of TBCs, particularly in scenarios involving small sample sizes facilitating assessments regarding their service life.

Surrogate model of turbulent transport in fusion plasmas using machine learning

Abstract The advent of machine learning (ML) has revolutionized the research of plasma confinement, offering new avenues for exploration. It enables the construction of models that effectively streamline the simulation process. While previous first-principles simulations have provided physics-based transport information, they have been inadequate fast for real-time applications or plasma control. In order to address this challenge, we introduce SExFC, a surrogate model based on the Gyro-Landau Extended Fluid Code (ExFC). An approach of physics-based database construction is detailed, as well the validity is illustrated. Through harnessing the power of ML, SExFC offers the capability to deliver rapid and precise predictions, facilitating real-time applications and enhancing plasma control. The proposed model integrates the recurrent neural network (RNN) algorithm, specifically leveraging the Gated Recurrent Unit (GRU) for iterative prediction of flux evolutions based on radial profiles. Therefore, the SExFC model has the potential to enable rapid and physics-based predictions that can be seamlessly integrated into future real-time plasma control systems.

Drug-associated glaucoma: A real-world study based on the Food and Drug Administration adverse event reporting system database.

This study aims to assess the risk of drug-associated glaucoma and track its epidemiological characteristics using real-world data. Adverse event reports from the Food and Drug Administration Adverse Event Reporting System (FAERS) from January 2004 to December 2023 were analysed. Disproportionality analysis and the Bayesian Confidence Propagation Neural Network algorithm were used. The study classified drugs associated with glaucoma, assessed risk levels, and compared drug-induced times across different categories. Eight hundred and five drugs were linked to glaucoma in the FAERS database. Disproportionality analysis identified 46 drugs with significant risk, mainly adrenergic medications (clobetasol propionate, fluocinolone acetonide), antihypertensives (hydrochlorothiazide), insulin (insulin human), anticholinergics (umeclidinium, darifenacin), VEGF inhibitors (brolucizumab, faricimab), and psychotropics (topiramate, ziprasidone). The top three high-risk drugs were clobetasol propionate, umeclidinium, and fluocinolone acetonide. The shortest drug-induced times were observed with indacaterol, salmeterol, and umeclidinium. Anticholinergic medications had the shortest drug-induced time among all categories. Females (62.5%) and the elderly (average age 63.5 ± 16.8 years) were predominantly affected. Reports of drug-associated glaucoma increased over the years. Preventing drug-associated glaucoma is more effective than treatment. Identifying the risk and drug-induced times of systemic and ophthalmic drugs can reduce occurrence risk. Clinical practitioners should be vigilant and inform patients of these risks.

Advances in Gas Detection of Pattern Recognition Algorithms for Chemiresistive Gas Sensor.

Gas detection and monitoring are critical to protect human health and safeguard the environment and ecosystems. Chemiresistive sensors are widely used in gas monitoring due to their ease of fabrication, high customizability, mechanical flexibility, and fast response time. However, with the rapid development of industrialization and technology, the main challenges faced by chemiresistive gas sensors are poor selectivity and insufficient anti-interference stability in complex application environments. In order to overcome these shortcomings of chemiresistive gas sensors, the pattern recognition method is emerging and is having a great impact in the field of sensing. In this review, we focus systematically on the advancements in the field of data processing methods for feature extraction, such as the methods of determining the characteristics of the original response curve, the curve fitting parameters, and the transform domain. Additionally, we emphasized the developments of traditional recognition algorithms and neural network algorithm in gas discrimination and analyzed the advantages through an extensive literature review. Lastly, we summarized the research on chemiresistive gas sensors and provided prospects for future development.

Visualization Analysis of Integrating College Sports Training and Psychology into Basketball Physical Education Teaching System Based on Image Recognition Algorithm

ABSTRACT On this basis, a basketball movement model based on ORB (Oriented FAST and Rotated BRIEF) local feature extraction has been proposed, and a basketball teaching visualization analysis system has been constructed. By extracting images from athletes’ videos in basketball, and using their training as a standard, tracking the frequency of their various celebration postures appearing and improving their scores in the game, corresponding feature points are extracted, and then the BP (back propagation) neural network algorithm is used to classify feature points. Ten different participants were selected from 10 games. Their movements were tracked, and then the extracted feature actions were imported into the visualization system for visual analysis, comparing the changes before and after psychological intervention. The results showed that there were significant differences in scores, errors, hit rates, steals, and motivation among subjects who received psychological cues and incentives compared to the data before psychological counseling. The scores and positive and negative values increased to 2–4 points, which had a certain positive impact on the competitive results.