Search

In this study, deep learning (DL) techniques are used for automatic recognition of soliton wave types. Accurate characterization of soliton wave species has the potential to improve their precise and effective use in various fields such as optics, electronics, telecommunications. In addition, the accuracy of the results obtained in the equation solutions will be demonstrated due to the determined wave type. Therefore, soliton analyses were performed at the beginning of the study using equations such as Korteweg-de Vries and nonlinear Schrödinger. These analyses led to the creation of 3D visual representations for eight distinct soliton types, including breather, kink, anti-kink, cusp, loop, lump, multi-peak, and rogue soliton. Following the generation of these images, we proceeded with a rigorous labeling process to prepare the data for the subsequent deep-learning phase. For this phase, we explored the performance of three prominent DL architectures: ResNet50V2, InceptionV3, and DenseNet169. Each architecture underwent separate training, validation, and testing procedures. Among these architectures, ResNet50V2 emerged as the top performer, consistently achieving high accuracies throughout the training, validation, and testing stages. Specifically, ResNet50V2 achieved training, validation, and testing accuracies of 0.9979, 1.00, and 1.00, respectively. Additionally, precision, recall, f1-score, weighted average, and macro average metrics all demonstrated perfect scores of 1.00. After completing the model training and evaluation process, we further assessed the model's performance by testing it on diverse 3D images, all of which resulted in predictions with 100% accuracy. Subsequently, we applied the ResNet50V2 architecture to test datasets representing six distinct soliton types documented in existing literature, successfully achieving accurate predictions for all instances. Through experiments conducted using both internally generated dataset pools and literature-derived images, the application of deep learning facilitated precise recognition of 3D soliton-type representations, underscoring its effectiveness in this domain. [ABSTRACT FROM AUTHOR]

With the rapid development of intelligent wearable technology, multimodal tactile sensors capable of data acquisition, decoupling of intermixed signals, and information processing have attracted increasing attention. Herein, a decoupled temperature–pressure dual‐mode sensor is developed based on single‐walled carbon nanotubes (SWCNT) and poly(3,4‐ethylenedioxythiophene): poly(styrenesulfonate) (PEDOT:PSS) decorated porous melamine foam (MF), integrating with a deep learning algorithm to obtain a multimodal input terminal. Importantly, the synergistic effect of PEDOT:PSS and SWCNT facilitates the sensor with ideal decoupling capability and sensitivity toward both temperature (38.2 µV K−1) and pressure (10.8% kPa−1) based on the thermoelectric and piezoresistive effects, respectively. Besides, the low thermal conductivity and excellent compressibility of MF also endow it with the merits of a low‐temperature detection limit (0.03 K), fast pressure response (120 ms), and long‐term stability. Benefiting from the outstanding sensing characteristics, the assembled sensor array showcases good capacity for identifying spatial distribution of temperature and pressure signals. With the assistance of a deep learning algorithm, it displays high recognition accuracy of 99% and 98% corresponding to "touch" and "press" actions, respectively, and realizes the encrypted transmission of information and accurate identification of random input sequences, providing a promising strategy for the design of high‐accuracy multimodal sensing platform in human–machine interaction. [ABSTRACT FROM AUTHOR]

With the continuous development of society, the universal applicability of Unmanned Aerial Vehicles (UAVs) has gradually emerged, because the facilities, such as surveillance cameras, lifting ropes and manipulators that can be modified are widely used in various fields, and are engaged in various dangerous work beyond the reach of human resources. It is not hard to imagine that in the near future, UAVs would inevitably join in more complex tasks. While UAVs have unique advantages, they also need to pay more attention to maintenance and care to ensure the stability of UAVs in the process of performing tasks and avoid the occurrence of a crash caused by imbalance. For this reason, this paper would set up an automatic balance control method based on Artificial Intelligence (AI) algorithm to study the balance of UAVs with manipulators. Through comparative study with the balance method based on deep learning algorithm, it was found that the method based on AI algorithm can help UAVs better maintain balance. The detection accuracy of the method studied in this paper for UAV flight balance stability was above 96%, and the detection accuracy of UAV automatic balance stability based on deep learning algorithm was below 92%. At the same time, in the face of different influencing factors, the UAV based on artificial intelligence algorithm can also maintain the balance of flight faster and maintain the stability of landing and development flight. Therefore, the research in this paper is meaningful. [ABSTRACT FROM AUTHOR]

Globally, monitoring productivity, occupational health, and safety of construction workers has long been a significant concern. To address this issue, there is an urgent need for efficient methods to continuously monitor construction sites and recognize workers' actions in a timely manner. Recently, advances in electronic technology and pose estimation algorithms have made it easier to obtain skeleton and joint trajectories of human bodies. Deep learning algorithms have emerged as robust and automated tools for extracting and processing 3D skeleton information on construction sites, proving effective for workforce action assessment. However, most previous studies on action recognition have primarily focused on single-stream data, which limited the network's ability to capture more comprehensive worker action features. Therefore, this research proposes a Spatial-Temporal Multi-Feature Network (STMF-Net) designed to utilize six 3D skeleton-based features to monitor and capture the movements of construction workers, thereby recognizing their actions. The experimental results demonstrate an accuracy of 79.36%. The significance of this work lies in its potential to enhance management models within the construction industry, ultimately improving workers' health and work efficiency. [ABSTRACT FROM AUTHOR]

Objectives: To examine the causal impact of small businesses on street theft and the underlying mechanisms. Methods: The "Cleanup Holes in the Wall" campaign in Beijing, China, provides a rare opportunity for a natural experiment. Drawing on street view images processed by deep learning algorithms and other big data sources such as court judgments and location-based service (LBS) population, we use difference-in-difference (DID) models to investigate how the disappearance of small businesses leads to changes in the occurrence of theft. We further examine the mechanisms by introducing mediators, including ambient population and social activity. Results: The treatment units that experienced a mass loss of small businesses showed a significant reduction in street theft compared to the control units that were less affected by the cleanup campaign. Ambient population and social activity played a mediating role in promoting and deterring crime, respectively, with the former dominating. The results remain robust after including covariates in the models, balancing covariates using the propensity score matching method, and adopting alternative thresholds to classify the treatment group. Conclusions: There are two competing yet coexisting mechanisms through which small businesses influence street theft. On the one hand, commercial premises provide large numbers of criminal opportunities for potential offenders; on the other hand, they are central to local social control and order. While small businesses exercise a certain amount of natural surveillance power, as a whole, they function primarily as crime generators. Implications for implementing targeted policies tailored to the nature of small businesses are discussed. [ABSTRACT FROM AUTHOR]

Accurate landslide displacement prediction is an essential prerequisite for early warning systems aimed at mitigating geological hazards. However, the inherent nonlinearity and dynamic complexity of landslide evolution often hinder forecasting performance. Previous studies have frequently combined signal decomposition techniques with individual machine learning methods to enhance prediction reliability. To address the limitations and uncertainties associated with individual models, this study presents a hybrid framework for displacement forecasting that combines variational mode decomposition (VMD) with multiple deep learning (DL) methods, including long short-term memory neural network (LSTM), gated recurrent unit neural network (GRU), and convolutional neural network (CNN), using a cloud model-based weighted strategy. Specifically, VMD decomposes cumulative displacement data into trend, periodic, and random components, thereby reducing the non-stationarity of raw data. Separate DL networks are trained to predict each component, and the forecasts are subsequently integrated through the cloud model-based combination strategy with optimally assigned weights. The proposed approach underwent thorough validation utilizing field monitoring data from the Baishuihe landslide in the Three Gorges Reservoir (TGR) region of China. Experimental results demonstrate the framework's capacity to effectively leverage the strengths of individual forecasting methods, achieving RMSE, MAPE, and R values of 12.63 mm, 0.46%, and 0.987 at site ZG118, and 20.50 mm, 0.52%, and 0.990 at site XD01, respectively. This combined approach substantially enhances prediction accuracy for landslides exhibiting step-like behavior. [ABSTRACT FROM AUTHOR]

Ground penetrating radar (GPR) is an effective tool for detecting internal defects in asphalt roads due to its non-destructive nature and high resolution. However, detecting defects in GPR images remains challenging, as existing models lack sufficient accuracy and are often complex and redundant. To address these issues, a lightweight real-time detection method based on ground-penetrating radar is proposed in this study. First, a field GPR image dataset of asphalt roads was collected and constructed. To address the limited defect sample data acquired by GPR, an efficient copy-and-paste augmentation method was employed. This method involved copying and pasting defect samples in GPR images while incorporating random scale jitter and position migration operations to generate a sufficient number of real defect samples. Second, the C2f-DSConv module and the SE attention mechanism were designed and introduced based on the YOLOv8 network to improve detection accuracy in the complex background environment of GPR images. Finally, a channel pruning strategy was used to prune the improved YOLOv8 network, reducing model complexity while maintaining detection accuracy. The final model achieves an average detection accuracy of 90.9% and a detection speed of 140.9 FPS. The results show that the proposed method combines both detection accuracy and real-time performance, further advancing the engineering application of internal defect detection in asphalt roads. [ABSTRACT FROM AUTHOR]

Highlights Filled rubber has wide applications in industries due to its high temperature and corrosion resistance. Therefore, it is crucial to accurately depict the high‐temperature mechanical behavior of the filled rubber. With the expansion of machine learning, the deep learning (DL) algorithm provides a new method to investigate the stress–strain relation of filled rubber. In this paper, the carbon nanotube‐filled fluororubber was used as an example to train various DL models, such as convolutional neural network (CNN), long short‐term memory (LSTM) network, and CNN‐LSTM hybrid models. These models were trained using test data at relatively lower temperatures to predict the relation between stress and strain at higher temperatures. Comparing the test results, it was found that all the predicted results closely matched the experimental data. However, the CNN‐LSTM hybrid model exhibited the lowest error and the most stable calculation process. The results indicated that the DL model not only reduces the time and resources needed to develop new constitutive relationships for filled rubber but also offers greater advantages in predicting the high‐temperature mechanical properties of filled rubber. DL models predict the high‐temperature mechanical behavior of filled rubber. Higher temperature results got by extrapolation from lower temperature data. More test data at various temperatures improve the prediction accuracy. The prediction accuracy at 200°C surpasses at 220°C. CNN‐LSTM model has the highest calculation efficiency and accuracy. [ABSTRACT FROM AUTHOR]

Introduction: Adolescent idiopathic scoliosis (AIS) is the most ordinary pediatric spinal disease. Timely intervention in growing individuals, such as brace treatment, relies on early detection of AIS. We developed a system consisting of a 3D depth sensor and an algorithm installed in a laptop computer. In this system, the correlation between the actual Cobb angle and the predicted Cobb angle calculated from the asymmetry index was 0.85 (P < 0.01). The purpose of this study is to create a deep learning algorithm (DLA) to identify moderate or severe AIS patients requiring the secondary screening using data of subjects detected in the school screening. Materials and Methods: We included 334 subjects detected using the 3D depth sensor system in school screening. The 3D images from the 3D depth sensor system were used as input data for the DLA with Convolutional neural networks. We randomly separated the 334 subjects into an internal validation data of 250 and an external validation data of 84. Binary classification was performed as 0 for images with Cobb angle of < 12° and 1 for images with Cobb angle of ≥ 12° based on the average actual Cobb angle of 12.0°. Five-fold cross validation was conducted to evaluate the probability for Cobb angle of ≥ 12°. The minimum predicted probability in subjects with Cobb angle of ≥ 15° was configured as the cut-off value to detect the second screening targets. In the external validation, 84 images were evaluated utilizing trained DLA in the internal validation, and decide to require secondary screening, based on the cu-off value. Results: In internal validation, the five-fold cross validation showed that the dataset 3 had the highest predicted performance. The minimum predicted probability in subjects with Cobb angle of ≥ 15° was 0.47 in dataset 3. In the external validation, the number of subjects with Cobb angle of < 10° and < 15° were 36 and 62, respectively. Based on a cut-off value of 0.47, 39 (63%) subjects with Cobb angle of < 15° were judged as unnecessary for the second screening. There was only one false negative case with Cobb angle of 19°. Conclusions: This DLA reduced the number of extremely mild AIS patient and false positive cases in the external validation, indicating that this DLA can reduce the unnecessary medical care expenditures and the unnecessary radiation exposure for children and adolescents. [ABSTRACT FROM AUTHOR]

Vannarut Satitpitakul,1,2 Apiwit Puangsricharern,3 Surachet Yuktiratna,3 Yossapon Jaisarn,4 Keeratika Sangsao,4 Vilavun Puangsricharern,1,2 Ngamjit Kasetsuwan,1,2 Usanee Reinprayoon,1,2 Thanachaporn Kittipibul1,2 1Center of Excellence for Cornea and Stem Cell Transplantation, Department of Ophthalmology, Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand; 2Excellence Center for Cornea and Stem Cell Transplantation, Department of Ophthalmology, King Chulalongkorn Memorial Hospital, Thai Red Cross Society, Bangkok, Thailand; 3IM Impower Company Limited, Bangkok, Thailand; 4Faculty of Medicine, Chulalongkorn University, Bangkok, ThailandCorrespondence: Vannarut Satitpitakul, Department of Ophthalmology, King Chulalongkorn Memorial Hospital, 1873 Rama 4 Road, Pathumwan, Bangkok, 10330, Thailand, Tel +66-894959022, Email Vannarut.S@chula.ac.thPurpose: To develop a comprehensively deep learning algorithm to differentiate between bacterial keratitis, fungal keratitis, non-infectious corneal lesions, and normal corneas.Methods: This retrospective study collected slit-lamp photos of patients with bacterial keratitis, fungal keratitis, non-infectious corneal lesions, and normal cornea. Causative organisms of infectious keratitis were identified by either positive culture or clinical response to single treatment. Convolutional neural networks (ResNet50, DenseNet121, VGG19) and Ensemble with probability weighting were used to develop a deep learning algorithm. The performance including accuracy, precision, recall, F1 score, specificity and AUC has been reported.Results: Total of 6478 photos from 2171 eyes, composed of 2400 bacterial keratitis, 1616 fungal keratitis, 1545 non-infectious corneal lesions, and 917 normal corneas were collected from hospital database. DenseNet121 demonstrated the best performance among three convolutional neural networks with the accuracy of 0.8 (95% CI 0.74– 0.86). The ensemble technique showed higher performance than single algorithm with the accuracy of 0.83 (95% 0.78– 0.88).Conclusion: Convolutional neural networks with ensemble techniques provided the best performance in discriminating bacterial keratitis, fungal keratitis, non-infectious corneal lesions, and normal corneas. Our models can be used as a screening tool for non-ophthalmic health care providers and ophthalmologists for rapid provisional diagnosis of infectious keratitis.Keywords: infectious keratitis, cornea ulcer, keratitis, conventional neural network, deep learning algorithm

WHO classified Helicobacter pylori as a Group I carcinogen for gastric cancer as early as 1994. However, despite the high prevalence of H. pylori infection, only about 3 % of infected individuals eventually develop gastric cancer, with the highly virulent H. pylori strains expressing cytotoxin-associated protein (CagA) and vacuolating cytotoxin (VacA) being critical factors in gastric carcinogenesis. It is well known that H. pylori infection is divided into two types in terms of the presence and absence of CagA and VacA toxins in serum, that is, carcinogenic Type I infection (CagA+/VacA+, CagA+/VacA-, CagA-/VacA+) and non-carcinogenic Type II infection (CagA-/VacA-). Currently, detecting the two carcinogenic toxins in active modes is mainly done by diagnosing their serological antibodies. However, the method is restricted by expensive reagents and intricate procedures. Therefore, establishing a rapid, accurate, and cost-effective way for serological profiling of carcinogenic H. pylori infection holds significant implications for effectively guiding H. pylori eradication and gastric cancer prevention. In this study, we developed a novel method by combining surface-enhanced Raman spectroscopy with the deep learning algorithm convolutional neural network to create a model for distinguishing between serum samples with Type I and Type II H. pylori infections. This method holds the potential to facilitate rapid screening of H. pylori infections with high risks of carcinogenesis at the population level, which can have long-term benefits in reducing gastric cancer incidence when used for guiding the eradication of H. pylori infections.

Abstract Weeds are undesired plants competing with crops for light, nutrients, and water, negatively impacting crop growth. Identifying weeds in wheat fields accurately is important for precise pesticide spraying and targeted weed control. Grass weeds in their early growth stages look very similar to wheat seedlings, making them difficult to identify. In this study, we focused on wheat fields with varying levels of grass weed infestation and used unmanned aerial vehicles (UAVs) to obtain images. By utilizing deep learning algorithms and spectral analysis technology, the weeds were identified and extracted accurately from wheat fields. Our results showed that the precision of weed detection in scattered wheat fields was 91.27% and 87.51% in drilled wheat fields. Compared to areas without weeds, the increase in weed density led to a decrease in wheat biomass, with the maximum biomass decreasing by 71%. The effect of weed density on yield was similar, with the maximum yield decreasing by 4320 kg·ha− 1, a drop of 60%. In this study, a method for monitoring weed occurrence in wheat fields was established, and the effects of weeds on wheat growth in different growth periods and weed densities were studied by accurately extracting weeds from wheat fields. The results can provide a reference for weed control and hazard assessment research.

The Brain Tumor Segmentation (BraTS) Challenge has been a main driver of the development of deep learning (DL) algorithms and provides by far the largest publicly available expert-annotated brain tumour dataset but contains solely preoperative examinations. The aim of our study was to facilitate the use of the BraTS dataset for training DL brain tumour segmentation algorithms for a postoperative setting. To this end, we introduced an automatic conversion of the three-label BraTS annotation protocol to a two-label annotation protocol suitable for postoperative brain tumour segmentation. To assess the viability of the label conversion, we trained a DL algorithm using both the three-label and the two-label annotation protocols. We assessed the models pre- and postoperatively and compared the performance with a state-of-the-art DL method. The DL algorithm trained using the BraTS three-label annotation misclassified parts of 10 out of 41 fluid-filled resection cavities in 72 postoperative glioblastoma MRIs, whereas the two-label model showed no such inaccuracies. The tumour segmentation performance of the two-label model both pre- and postoperatively was comparable to that of a state-of-the-art algorithm for tumour volumes larger than 1 cm3. Our study enables using the BraTS dataset as a basis for the training of DL algorithms for postoperative tumour segmentation.

As artificial intelligence technology continues to evolve, numerous advancements enable smoother communication and collaboration between humans and computers. Natural language processing (NLP), as a key direction in the field of computer and artificial intelligence, has been attracting much attention. It focuses on the language used by people in daily life, devoting itself to improving the ability of language generation and understanding, thus making communication between people and computers more fluent and natural, effectively breaking through the problem of human-computer interaction. With the advancement of deep learning technology, the migration of NLP models onto deep learning platforms has emerged as a key trend. Model transfer, a crucial aspect of deep learning, holds significant value in enhancing the performance and efficiency of NLP models. This paper begins by outlining the fundamentals of deep learning platforms and NLP model transfer, followed by a comprehensive examination of current research progress and challenges in this field. Subsequently, we introduce a novel NLP model transfer strategy tailored for deep learning platforms and validate its effectiveness through rigorous experiments. In conclusion, the paper highlights our noteworthy advancements, pointing toward promising future developments. [ABSTRACT FROM AUTHOR]

Flexible multimodal sensors have gained increasing popularity for applications in healthcare and extreme environment operations owing to their all‐around environmental perception and data acquisition capabilities. However, fabricating a magnetism‐mechanics‐humidity multimodal sensor that possesses high sensitivity without signal overlapping while in a facile methodology remains a great challenge. Herein, a highly sensitive, conductive, and flexible hydrogel sponge sensor with discriminable magnetism, mechanics, and humidity sensing capability is proposed, which shows stable pore size (19.30 µm) and satisfactory mechanical properties based on the synergistic hydrogen bonding among sodium alginate, poly(vinyl alcohol) and glycerol. The proposed sensors can not only display favorable humidity sensing ability with rapid response/recovery time (2.5/4 s) but also possess enhanced sensitivities (a gauge factor of 0.46 T−1 for magnetic field, −1.16 kPa−1 for pressure), superior stability and durability (over 8000 cycles). Benefiting from the separated capacitive and resistive response signals, the sensors can precisely distinguish the magnetic, mechanical, and humidity stimuli without cross‐talk. Further, the sensor arrays assisted by the deep learning algorithm are developed to realize gesture language recognition with a high accuracy of 99.17%. It can be believed that this high‐performance sensor will have good prospects in future soft electronics and human‐machine interaction systems. [ABSTRACT FROM AUTHOR]

Global warming worldwide is the reason for the increasing number of meteorological disasters causing severe property damage and human casualties. Railways are a key social infrastructure; however, quantitative and empirical research into the impact of weather changes due to global warming has not been done adequately. Thus, this study aims to develop a predictive model using a deep learning algorithm to quantify the relationship between fatal rail accidents and weather conditions. The proposed framework utilizes the Deep Neural Network (DNN) technique trained with past rail accidents and weather data. The model performance was evaluated using error metrics (mean absolute error (MAE) and root-mean-square error (RMSE)) and compared with widely used regression techniques. The findings showed that the DNN model achieved lower RMSE and MAE compared to the multi-regression, random forest and support vector machine models, with a reduction in prediction error ranging from 1.04% to 20.78% in RMSE and 5.0% to 15.3% in MAE. This exhibits the DNN model's effectiveness in capturing complex relationships within the data and delivering more accurate predictions compared to the other models. The approach and outcomes of this study provide essential guidelines for the efficient and safe maintenance and optimized safety management of railway services. [ABSTRACT FROM AUTHOR]

Weeds are undesired plants competing with crops for light, nutrients, and water, negatively impacting crop growth. Identifying weeds in wheat fields accurately is important for precise pesticide spraying and targeted weed control. Grass weeds in their early growth stages look very similar to wheat seedlings, making them difficult to identify. In this study, we focused on wheat fields with varying levels of grass weed infestation and used unmanned aerial vehicles (UAVs) to obtain images. By utilizing deep learning algorithms and spectral analysis technology, the weeds were identified and extracted accurately from wheat fields. Our results showed that the precision of weed detection in scattered wheat fields was 91.27% and 87.51% in drilled wheat fields. Compared to areas without weeds, the increase in weed density led to a decrease in wheat biomass, with the maximum biomass decreasing by 71%. The effect of weed density on yield was similar, with the maximum yield decreasing by 4320 kg·ha− 1, a drop of 60%. In this study, a method for monitoring weed occurrence in wheat fields was established, and the effects of weeds on wheat growth in different growth periods and weed densities were studied by accurately extracting weeds from wheat fields. The results can provide a reference for weed control and hazard assessment research. [ABSTRACT FROM AUTHOR]

The Brain Tumor Segmentation (BraTS) Challenge has been a main driver of the development of deep learning (DL) algorithms and provides by far the largest publicly available expert-annotated brain tumour dataset but contains solely preoperative examinations. The aim of our study was to facilitate the use of the BraTS dataset for training DL brain tumour segmentation algorithms for a postoperative setting. To this end, we introduced an automatic conversion of the three-label BraTS annotation protocol to a two-label annotation protocol suitable for postoperative brain tumour segmentation. To assess the viability of the label conversion, we trained a DL algorithm using both the three-label and the two-label annotation protocols. We assessed the models pre- and postoperatively and compared the performance with a state-of-the-art DL method. The DL algorithm trained using the BraTS three-label annotation misclassified parts of 10 out of 41 fluid-filled resection cavities in 72 postoperative glioblastoma MRIs, whereas the two-label model showed no such inaccuracies. The tumour segmentation performance of the two-label model both pre- and postoperatively was comparable to that of a state-of-the-art algorithm for tumour volumes larger than 1 cm3. Our study enables using the BraTS dataset as a basis for the training of DL algorithms for postoperative tumour segmentation. [ABSTRACT FROM AUTHOR]

With the rapid development of Internet technology, the amount of information we face is becoming more and more enormous. Traditional recognition methods have been unable to meet the needs of acquiring image content quickly and effectively. In this Internet age, developing websites is an essential job. From the perspective of deep learning algorithm, this paper studies and discusses a system based on deep neural network model, which aims to meet the user's personalized demand feature extraction. Through this system, the compatibility of users in different browsers can be obtained in real time, and the characteristics of images can be analyzed according to this information. This performance was then tested and analyzed, and the results showed that Chrome performed well in terms of compatibility, achieving the highest score of 9 out of all browsers. It was followed by Firefox and Safari, which scored 8 and 7 points, respectively. However, Edge and Opera had low compatibility scores of just 6 and 5, placing them at the bottom of the list. [ABSTRACT FROM AUTHOR]

Pinhole corrosions on oil and gas pipelines are difficult to detect and size and, therefore, pose a significant challenge to the pipeline integrity management practice. This study develops two convolutional neural network (CNN) models to identify pinholes and predict the sizes and location of the pinhole corrosions according to the magnetic flux leakage signals generated using the magneto-static finite element analysis. Extensive three-dimensional parametric finite element analysis cases are generated to train and validate the two CNN models. Additionally, comprehensive algorithm analysis evaluates the model performance, providing insights into the practical application of CNN models in pipeline integrity management. The proposed classification CNN model is shown to be highly accurate in classifying pinholes and pinhole-in-general corrosion defects. The proposed regression CNN model is shown to be highly accurate in predicting the location of the pinhole and obtain a reasonably high accuracy in estimating the depth and diameter of the pinhole, even in the presence of measurement noises. This study indicates the effectiveness of employing deep learning algorithms to enhance the integrity management practice of corroded pipelines. [ABSTRACT FROM AUTHOR]

Background: The rate at which the anticancer drug paclitaxel is cleared from the body markedly impacts its dosage and chemotherapy effectiveness. Importantly, paclitaxel clearance varies among individuals, primarily because of genetic polymorphisms. This metabolic variability arises from a nonlinear process that is influenced by multiple single nucleotide polymorphisms (SNPs). Conventional bioinformatics methods struggle to accurately analyze this complex process and, currently, there is no established efficient algorithm for investigating SNP interactions. Methods: We developed a novel machine‐learning approach called GEP‐CSIs data mining algorithm. This algorithm, an advanced version of GEP, uses linear algebra computations to handle discrete variables. The GEP‐CSI algorithm calculates a fitness function score based on paclitaxel clearance data and genetic polymorphisms in patients with nonsmall cell lung cancer. The data were divided into a primary set and a validation set for the analysis. Results: We identified and validated 1184 three‐SNP combinations that had the highest fitness function values. Notably, SERPINA1, ATF3 and EGF were found to indirectly influence paclitaxel clearance by coordinating the activity of genes previously reported to be significant in paclitaxel clearance. Particularly intriguing was the discovery of a combination of three SNPs in genes FLT1, EGF and MUC16. These SNPs‐related proteins were confirmed to interact with each other in the protein–protein interaction network, which formed the basis for further exploration of their functional roles and mechanisms. Conclusion: We successfully developed an effective deep‐learning algorithm tailored for the nuanced mining of SNP interactions, leveraging data on paclitaxel clearance and individual genetic polymorphisms. [ABSTRACT FROM AUTHOR]

Ting-Yu Hsu,1 Chi-Yung Cheng,1,2,&ast; I-Min Chiu,1,2 Chun-Hung Richard Lin,2 Fu-Jen Cheng,1 Hsiu-Yung Pan,1 Yu-Jih Su,3 Chao-Jui Li1,&ast; 1Department of Emergency Medicine, Kaohsiung Chang Gung Memorial Hospital, Chang Gung University College of Medicine, Kaohsiung, Taiwan; 2Department of Computer Science and Engineering, National Sun Yat-Sen University, Kaohsiung, Taiwan; 3Department of Internal Medicine, Division of Rheumatology, Allergy and Immunology, Kaohsiung Chang Gung Memorial Hospital, Kaohsiung, Taiwan&ast;These authors contributed equally to this workCorrespondence: Chi-Yung Cheng; Chao-Jui Li, Department of Emergency Medicine, Kaohsiung Chang Gung Memorial Hospital, 123 Ta-Pei Road, Niao-Sung, Kaohsiung, 833, Taiwan, Tel +886-7-7317123 ext. 8415, Email qzsecawsxd@gmail.com; chaojui@cgmh.org.twBackground: The global aging population presents a significant challenge, with older adults experiencing declining physical and cognitive abilities and increased vulnerability to chronic diseases and adverse health outcomes. This study aims to develop an interpretable deep learning (DL) model to predict adverse events in geriatric patients within 72 hours of hospitalization.Methods: The study used retrospective data (2017– 2020) from a major medical center in Taiwan. It included non-trauma geriatric patients who visited the emergency department and were admitted to the general ward. Data preprocessing involved collecting prognostic factors like vital signs, lab results, medical history, and clinical management. A deep feedforward neural network was developed, and performance was evaluated using accuracy, sensitivity, specificity, positive predictive value (PPV), and area under the receiver operating characteristic curve (AUC). Model interpretation utilized the Shapley Additive Explanation (SHAP) technique.Results: The analysis included 127,268 patients, with 2.6% experiencing imminent intensive care unit transfer, respiratory failure, or death during hospitalization. The DL model achieved AUCs of 0.86 and 0.84 in the validation and test sets, respectively, outperforming the Sequential Organ Failure Assessment (SOFA) score. Sensitivity and specificity values ranged from 0.79 to 0.81. The SHAP technique provided insights into feature importance and interactions.Conclusion: The developed DL model demonstrated high accuracy in predicting serious adverse events in geriatric patients within 72 hours of hospitalization. It outperformed the SOFA score and provided valuable insights into the model’s decision-making process. Keywords: explainable machine learning, deep learning algorithm, adverse events, mortality

A Block chain technology works well in a variety of problem domains is the bat-inspired algorithm (BA). Modified with Enhanced Bat Algorithm (M-EBA) transaction system to overcome the original BA's optimal limitation. In order to improve convergence speed to the optimal solution and avoid local optima trapping, it takes into account the frequency and velocity of the current best solution. The results show that the M-EBA algorithm performs noticeably better than competing algorithms. Additionally, M-EBA analyzes the effects of various parameters to identify the optimal combination of parameter values in the context of numerical optimization and successfully solves a real-world assignment problem which is typically solved using linear programming techniques, with satisfactory results. To detect the fraudulent transaction system in future directions for forthcoming developments are proposed, including the use of BA in large-scale, dynamic, robust, multi-objective optimization as as well as improve BA performance by utilizing election mechanisms, tuning parameters, structure population. [ABSTRACT FROM AUTHOR]

The efficiency of mechanical crushing is a key metric for evaluating machinery performance. However, traditional contact-based methods for measuring this efficiency are unable to provide real-time data monitoring and can potentially disrupt the production process. In this paper, we introduce a non-contact measurement technique for mechanical crushing efficiency based on deep learning algorithms. This technique utilizes close-range imaging equipment to capture images of crushed particles and employs deeply trained algorithmic programs rooted in symmetrical logical structures to extract statistical data on particle size. Additionally, we establish a relationship between particle size and crushing energy through experimental analysis, enabling the calculation of crushing efficiency data. Taking cement crushing equipment as an example, we apply this non-contact measurement technique to inspect cement particles of different sizes. Using deep learning algorithms, we automatically categorize and summarize the particle size ranges of cement particles. The results demonstrate that the crushing efficiencies of ore crushing particles, raw material crushing particles, and cement crushing particles can respectively reach 80.7%, 70.15%, and 80.27%, which exhibit a high degree of consistency with the rated value of the samples. The method proposed in this paper holds significant importance for energy efficiency monitoring in industries that require mechanical crushing. [ABSTRACT FROM AUTHOR]

In the domain of clinical imaging, the exact and quick identification proof of bone fractures plays a crucial part in a pivotal role in facilitating timely and effective patient care. This research tends to this basic need by harnessing the force of profound learning, explicitly utilizing a convolutional neural network (CNN) model as the foundation of our technique. The essential target of our study was to improve the mechanized recognition of bone fractures in X-ray images, utilizing the capacities of deep learning algorithms. The use of a CNN model permitted us to successfully capture and learn intricate patterns and features within the X-ray images, empowering the framework to make exact fracture detections. The training process included presenting the model to a various dataset, guaranteeing its versatility to an extensive variety of fracture types. The results of our research show the excellent performance of the CNN model in fracture detection, where our model has achieved an average precision of 89.5%, an average recall of 87%, and an overall accuracy of 91%. These metrics assert the vigour of our methodology and highlight the capability of deep learning in medical image analysis. [ABSTRACT FROM AUTHOR]

• Fusion of advanced deep-learning models CNN, LSTM, DNN and RNN with GA optimizes GWPZ mapping. • Overlaying GWPZ with drought maps unveils links between drought levels and groundwater potential. • CMIP6 climate models project changing GWPZ under different socioeconomic scenarios. • Workshops and local collaboration empower communities for sustainable water management. This study presents a novel approach to groundwater resource assessment and drought vulnerability, achieved through integrating cutting-edge deep-learning algorithms into a comprehensive framework. In regions susceptible to drought, ensuring groundwater availability is of paramount importance. Addressing this critical need, our research employs an ensemble of advanced algorithms, including long short-term memory, convolutional neural network, deep neural network, and recurrent neural network. These algorithms are further enhanced through optimization using a genetic algorithm to map groundwater potential zones (GWPZ). Leveraging model validation based on the area under the receiver operating characteristic curve (AUCROC), the long short-term memory-genetic algorithm model emerges as the superior algorithm, boasting the highest values of AUCROC: 0.995 for training and 0.996 for testing). Utilizing this optimized model, this study developed the GWPZ map, subsequently overlaying it with established drought maps developed by the Bangladesh Agricultural Research Council across distinct periods—Pre-Kharif, Kharif, and Rabi. The derived baseline GWPZ spatial distribution revealed five categories—very low (34.99%), low (27.67%), moderate (13.26%), high (11.71%), and very high (12.37%)—and their intersection with drought-prone regions, indicative of probabilities of drought occurrence ranging from very severe to low (15.14% to 24.69%). Moreover, employing the best-predicted model, this study projected future GWPZ for 2050 and 2100 using the coupled model intercomparison project 6 general circulation model data under the ambit of three distinct shared socioeconomic pathways (SSPs): SSP1-2.6, SSP2-4.5, and SSP5-8.5. Our findings suggested a contraction in the groundwater potential area by the end of the twenty-first century (2100). This pioneering integration sheds light on the intricate relationship between groundwater availability and drought susceptibility, furnishing invaluable insights for formulating targeted water resource management strategies. By amalgamating advanced computational techniques with geospatial analyses, this research contributes to a more comprehensive grasp of water resource dynamics within the context of escalating climate challenges. Consequently, it offers a foundation for informed decision-making and implementing sustainable water management practices in regions grappling with the dual challenges of groundwater scarcity and recurrent droughts. [ABSTRACT FROM AUTHOR]

In today's era of big data, people need to extract valuable and nutritious information from these data, but for the huge database, manual browsing and manual retrieval by ordinary users are obviously inefficient. Network data analysis can help users to study network data and deeply understand the embedded value of big data, which is popular in both civil and industrial and academic circles. This article aimed to use intelligent mathematical models to assist in the construction of network data analysis models and enabled the model to be repeatedly trained in the training set based on deep learning algorithms, becoming increasingly able to identify patterns between data. [ABSTRACT FROM AUTHOR]

Data protection issues stemming from the use of machine learning algorithms that are used in automated decision-making systems are discussed in this paper. More precisely, the main challenges in this area are presented, putting emphasis on how important it is to simultaneously ensure the accuracy of the algorithms as well as privacy and personal data protection for the individuals whose data are used for training the corresponding models. In this respect, we also discuss how specific well-known data protection attacks that can be mounted in processes based on such algorithms are associated with a lack of specific legal safeguards; to this end, the General Data Protection Regulation (GDPR) is used as the basis for our evaluation. In relation to these attacks, some important privacy-enhancing techniques in this field are also surveyed. Moreover, focusing explicitly on deep learning algorithms as a type of machine learning algorithm, we further elaborate on one such privacy-enhancing technique, namely, the application of differential privacy to the training dataset. In this respect, we present, through an extensive set of experiments, the main difficulties that occur if one needs to demonstrate that such a privacy-enhancing technique is, indeed, sufficient to mitigate all the risks for the fundamental rights of individuals. More precisely, although we manage—by the proper configuration of several algorithms’ parameters—to achieve accuracy at about 90% for specific privacy thresholds, it becomes evident that even these values for accuracy and privacy may be unacceptable if a deep learning algorithm is to be used for making decisions concerning individuals. The paper concludes with a discussion of the current challenges and future steps, both from a legal as well as from a technical perspective.

Globally, monitoring productivity, occupational health, and safety of construction workers has long been a significant concern. To address this issue, there is an urgent need for efficient methods to continuously monitor construction sites and recognize workers’ actions in a timely manner. Recently, advances in electronic technology and pose estimation algorithms have made it easier to obtain skeleton and joint trajectories of human bodies. Deep learning algorithms have emerged as robust and automated tools for extracting and processing 3D skeleton information on construction sites, proving effective for workforce action assessment. However, most previous studies on action recognition have primarily focused on single-stream data, which limited the network’s ability to capture more comprehensive worker action features. Therefore, this research proposes a Spatial-Temporal Multi-Feature Network (STMF-Net) designed to utilize six 3D skeleton-based features to monitor and capture the movements of construction workers, thereby recognizing their actions. The experimental results demonstrate an accuracy of 79.36%. The significance of this work lies in its potential to enhance management models within the construction industry, ultimately improving workers’ health and work efficiency.