Limited Accuracy Research Articles

Exploring the Premelting Transition through Molecular Simulations Powered by Neural Network Potentials

The premelting layer on crystal surfaces significantly affects the stability, surface reactivity, and phase transition behaviors of crystals. Traditional methods for studying this layer—experimental techniques, classical simulations, and even first-principle simulations—have significant limitations in accuracy and scalability. To overcome these challenges, we employ molecular dynamic simulations based on neural network potentials to investigate the structural and dynamic behavior of the premelting layer on ice. This approach matches the accuracy of first-principle calculations while greatly improving computational efficiency, allowing us to simulate the ice–vapor interface on a much larger scale. In this study, we conducted a one-nanosecond simulation of the ice–vapor interface involving 1024 water molecules. This significantly exceeds the time and size scales of previous first-principle studies. Our simulation results indicate complete surface melting. Furthermore, our simulation results reveal dynamic heterogeneity within the premelting layer, with molecules segregated into clusters of low and high mobility.

Development, validation and application of an LC-MS/MS method quantifying free forms of the micronutrients queuine and queuosine in human plasma using a surrogate matrix approach.

Queuosine (Q) is a hypermodified 7-deaza-guanosine nucleoside exclusively synthesized by bacteria. This micronutrient and its respective nucleobase form queuine (q) are salvaged by humans either from gut microflora or digested food. Depletion of Q-tRNA in human or mouse cells causes protein misfolding that triggers endoplasmic reticular stress and the activation of the unfolded protein responses. In vivo, this reduces the neuronal architecture of the mouse brain affecting learning and memory. Herein, a sensitive method for quantifying free q and Q in human blood was developed, optimised and validated. After evaluating q/Q extraction efficiency in several different solid-phase sorbents, Bond Elut PBA (phenylboronic acid) cartridges were found to have the highest extraction recovery for q (82%) and Q (71%) from pooled human plasma. PBS with 4% BSA was used as surrogate matrix for method development and validation. An LC-MS/MS method was validated across the concentration range of 0.0003-1 µM for both q and Q, showing excellent linearity (r2 = 0.997 (q) and r2 = 0.998 (Q)), limit of quantification (0.0003 µM), accuracy (100.39-125.71%) and precision (CV% < 15.68%). In a sampling of healthy volunteers (n = 44), there was no significant difference in q levels between male (n = 14; mean = 0.0068 µM) and female (n = 30; mean = 0.0080 µM) participants (p = 0.50). Q was not detected in human plasma. This validated method can now be used to further substantiate the role of q/Q in nutrition, physiology and pathology.

Enhancing the Resolution of Satellite Ocean Data Using Discretized Satellite Gridding Neural Networks

Ocean satellite data are often impeded by intrinsic limitations in resolution and accuracy. However, conventional data reconstruction approaches encounter substantial challenges when facing the nonlinear oceanic system and high-resolution fusion of variables. This research presents a Discrete Satellite Gridding Neural Network (DSGNN), a new machine learning method that processes satellite data within a discrete grid framework. By transforming the positional information of grid elements into a standardized vector format, the DSGNN significantly elevates the accuracy and resolution of data fusion through a neural network model. This method’s innovative aspect lies in its discretization and fusion technique, which not only enhances the spatial resolution of oceanic data but also, through the integration of multi-element datasets, better reflects the true physical state of the ocean. A comprehensive analysis of the reconstructed datasets indicates the DSGNN’s consistency and reliability across different seasons and oceanic regions, especially in its adept handling of complex nonlinear interactions and small-scale oceanic features. The DSGNN method has demonstrated exceptional competence in reconstructing global ocean datasets, maintaining small error variance, and achieving high congruence with in situ observations, which is almost equivalent to 1/12° hybrid coordinate ocean model (HYCOM) data. This study offers a novel and potent strategy for the high-resolution reconstruction and fusion of ocean satellite datasets.

A novel “S” chain structure mechanism model of magneto-rheological elastomer

Mechanism models based on chain-like structures are often used to characterize the morphology of magneto-rheological elastomer (MRE) and provide theoretical guidance for MRE preparation. However, widely used shear and tension mechanism models based on oblique straight chain or bent-chain structures have the limitation of insufficient accuracy in characterization, mainly because the internal particle chains of anisotropic MRE are not simple oblique straight or bent from scanning electron microscope (SEM) results. Therefore, the particle chain formation and particle interaction mechanism within MRE are revealed firstly. The theoretical results shows that particles tend to form S-like chain structures due to complex attraction and extrusion. On this basis, a novel “S” chain mechanism model is proposed, which is confirmed to have higher accuracy than the ideal shear model and tension model by comparing with the experimental results. The main reason is that the “S” chain mechanism model has realized the unification of shear, compressive, or tension mode, and integrates the fully coupled magnetic field, distribution parameters and interface stretching. The “S” chain mechanism model also takes into account the radius and volume fraction of the particles in MRE preparation, which makes it a more accurate guide to subsequent preparation.

Enhanced urban PM2.5 prediction: Applying quadtree division and time-series transformer with WRF-chem

Accurately predicting air quality levels in urban landscapes is a key environmental and public health challenge due to the complex interactions between pollutant emissions, atmospheric chemistry, and meteorological conditions. Chemical transport models such as Community Multiscale Air Quality (CMAQ) and Weather Research and Forecasting Modeling with Chemistry (WRF-Chem) provide fundamental insights into the behavior of atmospheric pollutants but face limitations in spatial resolution and prediction accuracy. Machine learning combined with low-cost sensors can improve prediction accuracy compared to numerical models, but uneven sensor distribution may result in biased prediction results. To address these limitations, this study introduces an innovative framework that leverages a quadtree space division to refine the numerical simulations of air pollutants into higher spatial resolutions by dynamically adjusting grid sizes based on the distribution of increased sensor deployments. Using a variable-resolution grid, the study first aggregates variables into a grid and then calculates spatial dependence based on grid cells. The deep-learning-based Time Series Transformer model is then used to perform detailed temporal predictions of PM2.5 concentration across the entire grid. The spatial and temporal modeling framework is designed to use the past 48 h' aggregated data, including meteorological variables from WRF-Chem numerical simulation and PM2.5 concentrations from ground monitoring stations and sensors, to predict future 48-h’ PM2.5 concentrations and tested across entire grids within Los Angeles County from January to June 2020. The results show that the performance of variable resolution grids is better than that of fixed grids regarding four metrics: minimum resolution, overall spatiotemporal RMSE, average bias, and prediction coverage. Among them, the variable resolution grids with up to 4 sensors showed the best performance, with an overall spatiotemporal RMSE of 1.12 μg/m3, which was about 40% lower than 2-sensor and 8-sensor variable resolution grids, a significant reduction of 55% compared to a 3 km fixed grid configuration. Future work will involve additional relevant factors such as emission sources and sinks to improve prediction accuracy.

Determination and contamination status of mercury and other heavy metals in artisanal and small-scale gold mining at Banyumas region, Central Java, Indonesia, using ICP-OES

ABSTRACT The determination of mercury and other heavy metals (As, Cd, Cr, Ni, and Pb) has been carried out as toxic metals for the human body. Sample preparation was carried out by the microwave digested assist. The determination of heavy metals was carried out using the ICP-OES. The validation parameters used were linearity, Limit of Detection (LoD), Limit of Quantitation (LoQ), precision, and accuracy. Mercury, Arsen, Cadmium, Chromium, Nickel, and Lead average content in sediment gold ore processing waste was 1617.64, 5525.09, 20.24, 484.96, 259.67, and 1210.27 μg/kg, respectively. Linearity used a standard solution of 10–1000 μg/kg, and it produced a regression equation with r2 > 0.9992. The Limit of Detection (LoD) and the Limit of Quantitation (LoQ) was below 1.56 μg/kg and 6.27 μg/kg, respectively. The precision was calculated by the repeatability method, with a result of %RSD below 2.30%. Accuracy was obtained by recovering a spiked standard. The recovery value was found to be between 83% and 118%. Geoaccumulation Index (Igeo), Metal Pollution Index (MPI), Degree of Contamination (Cdeg), and Potential Ecological Risk Index (PERI) in this study was −8.62–1.43, 0.02–5.33, 8.80, and 168.59, respectively. The waste processing site at Artisanal and Small-Scale Gold Mining is generally contaminated by mercury and other heavy metals. All the validation parameters of the method are in the acceptability range, so the method can be used as a routine method for determining mercury and other heavy metals (As, Cd, Cr, Ni, and Pb) levels in gold ore processing waste.

Quantizing Galilean spacetime: a reconstruction of Maxwell’s equations in empty space

As was recently shown, non-relativistic quantum theory can be derived by means of a projection method from a continuum of classical solutions for (massive) particles. In this paper, we show that Maxwell’s equations in empty space can be derived using the same method. In this case, the starting point is a continuum of solutions of equations of motion for massless particles describing the structure of Galilean space-time. As a result of the projection, the space-time structure itself is changed by the appearance of a new fundamental constant c with the dimension of a velocity. This maximum velocity c, derived here for massless particles, is analogous to the accuracy limit ħ\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\hbar $$\\end{document} derived earlier for massive particles. The projection method can thus be interpreted as a generalized quantization. We suspect that all fundamental fields can be traced back to continuous sets of particle trajectories, and that in this sense, the particle concept is more fundamental than the field concept.

Harnessing the Potential of Artificial Intelligence in Managing Viral Hepatitis

Viral hepatitis continues to be a serious global health concern, impacting millions of people, putting a strain on healthcare systems across the world, and causing significant morbidity and mortality. Traditional diagnostic, prognostic, and therapeutic procedures to address viral hepatitis are successful but have limits in accuracy, speed, and accessibility. Artificial intelligence (AI) advancement provides substantial opportunities to overcome these challenges. This study investigates the role of AI in revolutionizing viral hepatitis care, from early detection to therapy optimization and epidemiological surveillance. A comprehensive literature review was conducted using predefined keywords in the Nature, PLOS ONE, PubMed, Frontiers, Wiley Online Library, BMC, Taylor &amp; Francis, Springer, ScienceDirect, MDPI, IEEE Xplore Digital Library, and Google Scholar databases. Peer-reviewed publications written in English between January 2019 and August 2024 were examined. The data of the selected research papers were synthesized and analyzed using thematic and narrative analysis techniques. The use of AI-driven algorithms in viral hepatitis control involves many significant aspects. AI improves diagnostic accuracy by integrating machine learning (ML) models with serological, genomic, and imaging data. It enables tailored treatment plans by assessing patient-specific characteristics and predicting therapy responses. AI-powered technologies aid in epidemiological modeling, and AI-powered systems effectively track treatment adherence, identify medication resistance, and control complications associated with chronic hepatitis infections. It is vital in identifying new antiviral medicines and vaccines, speeding the development pipeline through high-throughput screening and predictive modeling. Despite its transformational promise, using AI in viral hepatitis care presents various challenges, including data privacy concerns, the necessity for extensive and varied datasets, and the possibility of algorithmic biases. Ethical considerations, legal frameworks, and multidisciplinary collaboration are required to resolve these issues and ensure AI technology’s safe and successful use in clinical practice. Exploiting the full AI’s potential for viral hepatitis management provides unparalleled prospects to improve patient outcomes, optimize public health policies, and, eventually, and alleviate the disease’s negative impact worldwide. This study seeks to provide academics, medics, and policymakers with the fundamental knowledge they need to harness AI’s potential in the fight against viral hepatitis.

Prospect of large language models and natural language processing for lung cancer diagnosis: A systematic review

AbstractLung cancer, a leading cause of global mortality, demands a combat for its effective prevention, early diagnosis, and advanced treatment methods. Traditional diagnostic methods face limitations in accuracy and efficiency, necessitating innovative solutions. Large Language Models (LLMs) and Natural Language Processing (NLP) offer promising avenues for overcoming these challenges by providing comprehensive insights into medical data and personalizing treatment plans. This systematic review explores the transformative potential of LLMs and NLP in automating lung cancer diagnosis. It evaluates their applications, particularly in medical imaging and the interpretation of complex medical data, and assesses achievements and associated challenges. Emphasizing the critical role of Artificial Intelligence (AI) in medical imaging, the review highlights advancements in lung cancer screening and deep learning approaches. Furthermore, it underscores the importance of on‐going advancements in diagnostic methods and encourages further exploration in this field.

Data-based modeling of the Pharmacodynamics for the effect of Propofol and Remifentanil during General Anesthesia

Predicting Bispectral Index (BIS) and Mean Arterial Pressure (MAP) during anesthesia is critical for patient safety and effective anesthesia management. Traditional pharmacodynamic response surface models have limitations in accuracy and adaptability. This paper presents a novel approach for predicting BIS and MAP using machine learning techniques. Rather than using standard pharmacodynamic response surface models, a machine learning-based approach is proposed to model the pharmacodynamic. The proposed method considers the state of the standard propofol and remifentanil pharmacokinetic model and the patient information as features to predict the BIS and MAP values. Training and testing are done on a selected subset of the VitalDB dataset (Lee and Jung, 2018) containing 191 different patients. Results demonstrate that the machine learning-based approach outperforms standard pharmacodynamic models in terms of accuracy. Specifically, the Support Vector Regression (SVR) model achieves a Mean Absolute Prediction Error (MDAPE) 32% smaller than the Eleveld model for BIS prediction. For MAP prediction, the SVR model also demonstrates superior performance with a reduction of 66% of MDAPE. The proposed method provides similar performance as the deep-learning method (Lee et al., 2018) while keeping a simple structure that may be used in other applications.

Microbial cell-free DNA-sequencing as an addition to conventional diagnostics in neonatal sepsis.

Bloodstream infections remain a challenge for neonatologists, as traditional culture-based methods are time-consuming and rely on adequate blood volume. Next-generation sequencing (NGS) offers an alternative, as it can identify microbial cell-free DNA (mcfDNA) in a small blood sample, providing rapid pathogen detection. This study aimed to assess the diagnostic performance of DISQVER®-NGS compared to blood cultures in neonatal patients with suspected sepsis. In neonates with suspected sepsis, blood cultures and samples for NGS were prospectively collected. Patients were divided into four categories: 1) sepsis, blood culture positive, 2) clinical sepsis, culture negative, 3) suspected sepsis, 4) validation cohort. NGS detected bacterial, viral or fungal mcfDNA in 24 of 82 samples. Blood cultures were collected in 46 of 84 patients (15/46 positive). DISQVER® correctly identified pathogens in 9/15 patients with a positive blood culture, two with intrinsic resistance to their antibiotic regimen. In seven samples NGS reported the mcfDNA of bacteria that could have theoretically grown in culture but did not. NGS may enhance sensitivity in sepsis diagnostics by detecting mcfDNA in neonates with suspected sepsis. Interpreting NGS results requires correlation with clinical data, laboratory values, and routine microbiological tests for a comprehensive understanding of the patient's condition. Conventional blood culture methods have limitations in accuracy and turnaround time. The study aimed to investigate the diagnostic performance of the Next-Generation Sequencing method DISQVER® compared to traditional blood cultures in neonatal patients with suspected sepsis. Our findings suggest that NGS has the potential to augment the precision of conventional diagnostic techniques, can lead to improved detection of pathogens and targeted treatment approaches in neonatal sepsis. It is emphasized that further validation and integration with clinical and microbiological data are required to ensure optimal clinical utility.

Deep Learning Method for Real-Time Fire Detection System for Urban Fire Monitoring and Control

During urban fire incidents, real-time videos and images are vital for emergency responders and decision-makers, facilitating efficient decision-making and resource allocation in smart city fire monitoring systems. However, real-time videos and images require simple and embeddable models in small computer systems with highly accurate fire detection ratios. YOLOv5s has a relatively small model size and fast processing time with limited accuracy. The aim of this study is to propose a method that employs a YOLOv5s network with a squeeze-and-excitation module for image filtering and classification to meet the urgent need for rapid and accurate real-time screening of irrelevant data. In this study, over 3000 internet images were used for crawling and annotating to construct a dataset. Furthermore, the YOLOv5, YOLOv5x and YOLOv5s models were developed to train and test the dataset. Comparative analysis revealed that the proposed YOLOv5s model achieved 98.2% accuracy, 92.5% recall, and 95.4% average accuracy, with a remarkable processing speed of 0.009 s per image and 0.19 s for a 35 frames-per-second video. This surpasses the performance of other models, demonstrating the efficacy of the proposed YOLOv5s for real-time screening and classification in smart city fire monitoring systems.

Predicting ultimate strength and axial strain of circular concrete columns having AFRP wraps and tube-encased confinement

The current study performs an in-depth analysis into the reinforcement of circular concrete columns (CCC) using aramid fiber-reinforced polymers (AFRP), focusing on two distinct techniques: wet-layup wrapping and tube-encasement with manufactured shells. The need for improved forecasting models for AFRP-wrapped concrete structures arises from observed inconsistencies and limited accuracy in existing models, which often fail to fully account for key parameters such as confining stiffness and composite strain. Addressing this gap, our study begins with a comprehensive review of prior experimental research, emphasizing the impact of various parameters on the performance of AFRP-wrapped CCC under monotonic axial compressive loading. To build a robust foundation for this investigation, we compiled an extensive database of experimental outcomes from existing literature, encompassing 180 concrete cylinders wrapped by AFRP. Our analysis meticulously evaluates nine different models designed to predict the ultimate compressive strength and axial strain of circular columns reinforced with AFRP. These models are scrutinized for their accuracy, revealing notable discrepancies, particularly when key parameters are overlooked. In response, we introduce new equations that offer more precise predictions compared to previously proposed models. These novel equations are derived from an extensive statistical analysis, ensuring their reliability and robustness. Prominently, our proposed model for compressive strength demonstrated superior performance with a coefficient of determination (R2 = 0.87), a root means square error (RMSE) of 0.29, and a mean absolute error (MAE) of 22.91. Similarly, the proposed strain model showed a high level of forecasting accuracy, with a coefficient of determination (R2 = 0.82), a root mean square error (RMSE) of 2.79, and a mean absolute error (MAE) of 208.69. To validate our findings, we employed the Taylor diagram and probability density function, which provide a detailed comparison of the forecasting accuracy of our proposed models against existing ones. The outcomes of this study not only enhance the understanding of AFRP-wrapped concrete columns but also offer valuable insights for the development of more effective reinforcement techniques. Our findings are expected to contribute significantly to the advancement of structural engineering practices, particularly in the application of AFRP for strengthening concrete structures.

At the Limit of Interfacial Sharpness in Nanowire Axial Heterostructures.

As semiconductor devices approach dimensions at the atomic scale, controlling the compositional grading across heterointerfaces becomes paramount. Particularly in nanowire axial heterostructures, which are promising for a broad spectrum of nanotechnology applications, the achievement of sharp heterointerfaces has been challenging owing to peculiarities of the commonly used vapor-liquid-solid growth mode. Here, the grading of Al across GaAs/AlxGa1-xAs/GaAs heterostructures in self-catalyzed nanowires is studied, aiming at finding the limits of the interfacial sharpness for this technologically versatile material system. A pulsed growth mode ensures precise control of the growth mechanisms even at low temperatures, while a semiempirical thermodynamic model is derived to fit the experimental Al-content profiles and quantitatively describe the dependences of the interfacial sharpness on the growth temperature, the nanowire radius, and the Al content. Finally, symmetrical Al profiles with interfacial widths of 2-3 atomic planes, at the limit of the measurement accuracy, are obtained, outperforming even equivalent thin-film heterostructures. The proposed method enables the development of advanced heterostructure schemes for a more effective utilization of the nanowire platform; moreover, it is considered expandable to other material systems and nanostructure types.

Analysis of heavy metals and minerals in edible vegetable oils produced and marketed in Gondar City, Northwest Ethiopia

IntroductionNowadays, food safety is regarded as one of the most critical global public health issues. Edible oil, a key ingredient in food processing, is widely used and consumed in every Ethiopian household. However, its safety is often overlooked. Currently, edible oil is produced in Ethiopia from small-scale operations to large industrial levels, as well as imported from other countries.ObjectiveThis study aimed to determine the levels of heavy metals and essential minerals in edible vegetable oils produced and marketed in Gondar City, Northwest Ethiopia.MethodsA laboratory-based cross-sectional study was conducted from May to July 2021 in Gondar City. Seventeen edible oil samples were collected using simple random sampling techniques. Heavy metal content was determined using an atomic absorption spectrophotometer with the standard procedures and techniques after microwave digestion. The efficiency and validity of the method used were evaluated by determining the limit of detection (LOD), the limit of quantification (LOQ), accuracy, and precision. The collected data were entered into Microsoft Excel and transported to Stata for analysis.ResultA total of seventeen vegetable oil samples were analyzed. The accuracy of the method was evaluated by recovery studies, which ranged from 81 to 115%, and the relative standard deviations were found to be below 15%. The concentrations of Zn, Cu, Fe, Cd, and Pb were in the range of 0.07 to 0.8 mg/l, 0.002 to 0.06 mg/l, 0.01 to 0.8 mg/l, 0.08 to 0.18 mg/l, and 0.003 to 0.27 mg/l, respectively. In general, the lead and cadmium content was higher than other metals in some of the investigated edible vegetable oils. Most values fell within the permissible quality limits for edibility as prescribed by the World Health Organization (WHO) and the National Agency for Food and Drug Administration and Control (NAFDAC). However, the levels of Pb and Cd exceeded the reference levels in some locally produced vegetable oils.ConclusionTo address the exceeded levels of heavy metals, it is imperative to implement more careful handling, processing of raw materials, and filtering practices. Producers and marketers should take the necessary precautions to prevent contamination. Strict regulatory control from responsible bodies and stakeholders is recommended to ensure the safety and metal contents of vegetable oils originating from the study area.

What controls hydrology? An assessment across the contiguous United States through an interpretable machine learning approach

Machine learning (ML) is increasingly perceived as a futuristic, superior data-driven approach to scientific discovery. It has already demonstrated remarkable performance in forecasting and prediction, yet its role in enhancing our understanding of hydrological processes remains underexplored. Traditional hydrological interpretations have relied heavily on model-dependent interpretation methods, focusing on the predictive accuracy of ML model predictions. Since hydrological models are built on a collection of assumptions and simplifications, model-dependent approaches might suffer from limited model realism, adequacy, accuracy, and equifinality issues. To address this gap, this study provides an ML approach that works in a model-independent context, working directly on hydroclimatic data collected through monitoring systems.We apply our model-independent interpretation approach to a carefully designed set of hydrologic data collected across the contiguous United States to address the following questions: (1) What are the primary controls of runoff-generation mechanisms, and how can such controls be attributed to catchment properties? (2) How and under what circumstances can the history of climate variables, such as precipitation, be a surrogate for present-time state variables, such as soil moisture and snowpack? We show that the ML approach aids in distinguishing catchments characterized by strong overland flow, interflow, or baseflow components and those primarily driven by rainfall, snowmelt, or a mix thereof. We further show that typical surrogate variables used in hydrology may come short in representing the dynamics of catchments that exhibit a complex interplay of rain and snow.

Geometric deep learning for molecular property predictions with chemical accuracy across chemical space

Chemical engineers heavily rely on precise knowledge of physicochemical properties to model chemical processes. Despite the growing popularity of deep learning, it is only rarely applied for property prediction due to data scarcity and limited accuracy for compounds in industrially-relevant areas of the chemical space. Herein, we present a geometric deep learning framework for predicting gas- and liquid-phase properties based on novel quantum chemical datasets comprising 124,000 molecules. Our findings reveal that the necessity for quantum-chemical information in deep learning models varies significantly depending on the modeled physicochemical property. Specifically, our top-performing geometric model meets the most stringent criteria for “chemically accurate” thermochemistry predictions. We also show that by carefully selecting the appropriate model featurization and evaluating prediction uncertainties, the reliability of the predictions can be strongly enhanced. These insights represent a crucial step towards establishing deep learning as the standard property prediction workflow in both industry and academia.Scientific contributionWe propose a flexible property prediction tool that can handle two-dimensional and three-dimensional molecular information. A thermochemistry prediction methodology that achieves high-level quantum chemistry accuracy for a broad application range is presented. Trained deep learning models and large novel molecular databases of real-world molecules are provided to offer a directly usable and fast property prediction solution to practitioners.

Deep learning-based quality prediction for multi-stage sequential hot rolling processes in heavy rail manufacturing

Traditional quality control practices in heavy rail manufacturing primarily rely on experiential knowledge and numerical simulations. However, these methods come with significant drawbacks, including high costs, safety concerns, and limited accuracy due to the inadequate consideration of various influencing factors. Despite the accumulation of extensive data by some heavy rail manufacturers in recent years, achieving data-driven quality modeling remains a formidable challenge, mainly due to the complexity of the multi-stage sequential production process. This study presents the Temporal-Spatial Mapping approach to transform isochronous sampling data into spatially aligned data. Additionally, it introduces a multi-objective prediction model, Temporal-Fusion-LSTM (TFL), designed for evaluating the quality of heavy rails. The proposed model presents the Double-Layer LSTM with Time-Steps Fusion (DLL) architecture. The first layer of the DLL architecture targets the temporal transfer regularity within a single stage, while the second layer addresses the transition dynamics across multiple rolling stages in the heavy rail rolling process. Following this, the Local Process Correlation Matrix (LPCM) based on the Maximal Information Coefficient (MIC) is employed to empower the model to comprehend the intricate interactions among parameters in each stage of the rolling process. Finally, the Multi-Gate based Shared-Bottom architecture is employed to accentuate manufacturing processes highly correlated with the target quality indicators during multi-objective prediction. Real production data from a heavy rail factory is utilized to validate the effectiveness of the method. The experimental results unequivocally demonstrate that the proposed method accurately predicts heavy rail quality and facilitates cross-stage prediction, thus establishing the groundwork for data-driven quality diagnosis and pre-control in heavy rail production.

DCASAM: advancing aspect-based sentiment analysis through a deep context-aware sentiment analysis model

Aspect-level sentiment analysis plays a pivotal role in fine-grained sentiment categorization, especially given the rapid expansion of online information. Traditional methods often struggle with accurately determining sentiment polarity when faced with implicit or ambiguous data, leading to limited accuracy and context-awareness. To address these challenges, we propose the Deep Context-Aware Sentiment Analysis Model (DCASAM). This model integrates the capabilities of Deep Bidirectional Long Short-Term Memory Network (DBiLSTM) and Densely Connected Graph Convolutional Network (DGCN), enhancing the ability to capture long-distance dependencies and subtle contextual variations.The DBiLSTM component effectively captures sequential dependencies, while the DGCN component leverages densely connected structures to model intricate relationships within the data. This combination allows DCASAM to maintain a high level of contextual understanding and sentiment detection accuracy.Experimental evaluations on well-known public datasets, including Restaurant14, Laptop14, and Twitter, demonstrate the superior performance of DCASAM over existing models. Our model achieves an average improvement in accuracy by 1.07% and F1 score by 1.68%, showcasing its robustness and efficacy in handling complex sentiment analysis tasks.These results highlight the potential of DCASAM for real-world applications, offering a solid foundation for future research in aspect-level sentiment analysis. By providing a more nuanced understanding of sentiment, our model contributes significantly to the advancement of fine-grained sentiment analysis techniques.

A piecewise extreme learning machine for interface problems

Deep learning methods have been developed to solve interface problems, benefiting from meshless features and the ability to approximate complex interfaces. However, existing deep neural network (DNN) methods for usual partial differential equations encounter accuracy limitations where after reaching a certain error level, further increases in network width, depth, and iteration steps do not enhance accuracy. This limitation becomes more notable in interface problems where the solution and its gradients may exhibit significant jumps across the interface. To improve accuracy, we propose a piecewise extreme learning machine (PELM) for addressing interface problems. An ELM is a type of shallow neural network where weight/bias coefficients in activation functions are randomly sampled and then fixed instead of being updated during the training process. Considering the solution jumps across the interface, we use a PELM scheme — setting one ELM function for each side of the interface. The two ELM functions are coupled using the interface conditions. Our numerical experiments demonstrate that the proposed PELM for the interface problem significantly improves the accuracy compared to conventional DNN solvers. The advantage of new method is shown for addressing interface problems that feature complex interface curves.