Network Model Research Articles

Cryptocurrency price fluctuation and time series analysis through candlestick pattern of bitcoin and ethereum using machine learning

PurposeCandlestick charts are a key tool for the technical analysis of cryptocurrency price fluctuations. It is essential to examine trends in the time series of a financial asset when completing an analysis. To accurately examine its potential future performance, it must also consider how it has changed and been active during the period. The researchers created cryptocurrency trading algorithms in this study based on the traditional candlestick pattern.Design/methodology/approachThe data includes information on Bitcoin prices from early 2012 until 2021. Only the engulfing Candlestick model was able to anticipate changes in the price movements of Bitcoin. The traditional Harami model does not work with Bitcoin trading platforms because it has yet to generate profitable business results. An inverted Harami is a successful cryptocurrency trading method.FindingsThe inverted Harami approach accounts for 6.98 profit factor (PrF) and 74–50% of profitable (Pr) transactions, which favors a particularly long position. Additionally, the study discovered that almost all analyzed candlestick patterns forecast longer trends greater than shorter trends.Research limitations/implicationsTo statistically study its future potential return, examining how it has changed and been active over the years is necessary. Such valuations are the basis for trading strategies that could help traders and investors in the cryptocurrency market. Without sacrificing clarity or ease of application, the proposed approach has increased performance by up to 32.5% of mean absolute error (MAE).Originality/valueThis study is novel in that it used multilayer autoregressive neural network (MARN) models with crypto-net (CNM) in machine learning to analyze a time series of financial cryptocurrencies. Here, the primary study deals with time trends extracted through a neural network model. Then, the developed model was tested using Bitcoin and Ethereum. Finally, CNM validity was tested through linear regression.

Confidence intervals for mean response times of a open queueing network: calibration approach

Response time is a key factor in the analysis of the different queueing network model properties. This study computes a series of response times using a data-based recurrence relation. The true average response time is determined using the sample averages from those response times. Several confidence intervals are created for the open queueing net- work model’s average response times using the calibration approach. Using a numerical simulation analysis, the accuracy of the various confidence intervals is evaluated. Journal of Statistical Research 2023, Vol 57, No.1-2, pp.69-79

Regression modeling and multi-objective optimization of rheological behavior of non-Newtonian hybrid antifreeze: Using different neural networks and evolutionary algorithms

The research used an artificial neural network (ANN) model to examine the rheological properties of hybrid non-Newtonian ferrofluids (HNFFs) composed of Fe-CuO, water, and ethylene glycol. The performance of neural network was optimized using seven regression methods (RMs), namely Group Method of Data Handling (GMDH), Decision Tree (D-Tree), Multi-Layer Perceptron (MLP), Support Vector Machine (SVM), Extreme Learning Machine (ELM), Radial Basis Function (RBF), and Multiple Linear Regression (MLR). The findings highlighted GMDH method's superior performance when compared to neural networks. R and RMSE values attained by GMDH for the objective function (OF) μnf were 0.99436 and 2.0135, respectively. For the torque function OF, the values were 0.97652 and 4.8952. Margin of difference (MOD) calculations across various algorithms, such as MLP, SVM, RBF, D-Tree, ELM, MLR, and GMDH-Algos revealed significant disparities, indicating GMDH's efficacy. Comparison of R, RMSD, and standard deviation values between GMDH and MLR algorithms further underscored performance discrepancies. Specific parameters for which NSGA II Algo was rated highest among evaluation indices were as follows: a crossover rate of 0.7, a mutation rate of 0.02, a population size of 50, and 500 generations. Post-optimization, optimal values for μnf and torque (To) were determined as 6.595 and 3.543, respectively, with corresponding values for φ, T, and γ obtained as 0.185, 49.372, and 3.163, respectively. This comprehensive analysis sheds light on the effectiveness of various regression methods in modeling the rheological behavior of hybrid non-Newtonian ferrofluids, contributing to advancements in fluid dynamics research.

TinyM 2 Net-V2: A Compact Low-power Software Hardware Architecture for M ulti m odal Deep Neural Networks

With the evaluation of Artificial Intelligence (AI), there has been a resurgence of interest in how to use AI algorithms on low-power embedded systems to broaden potential use cases of the Internet of Things (IoT). To mimic multimodal human perception, multimodal deep neural networks (M-DNN) have recently become very popular with the classification task due to their impressive performance for computer vision and audio processing tasks. This article presents TinyM 2 Net-V2 —a compact low-power software hardware architecture for m ulti m odal deep neural networks for resource-constrained tiny devices. To compress the models to implement on tiny devices, cyclicly sparsification and hybrid quantization (4-bits weights and 8-bits activations) methods are used. Although model compression techniques are an active research area, we are the first to demonstrate their efficacy for multimodal deep neural networks, using cyclicly sparsification and hybrid quantization of weights/activations. TinyM 2 Net-V2 shows that even a tiny multimodal deep neural network model can improve the classification accuracy more than that of any unimodal counterparts. Parameterized M-DNN model architecture was designed to be evaluated in two different case-studies: vehicle detection from multimodal images and audios and COVID-19 detection from multimodal audio recordings. The most compressed TinyM 2 Net-V2 achieves 92.5% COVID-19 detection accuracy (6.8% improvement from the unimodal full precision model) and 90.6% vehicle classification accuracy (7.7% improvement from the unimodal full precision model). A parameterized and flexible FPGA hardware accelerator was designed as well for TinyM 2 Net-V2 models. To the best of our knowledge, this is the first work accelerating multimodal deep neural network models on low-power Artix-7 FPGA hardware. We achieved energy efficiency of 9.04 GOP/s/W and 15.38 GOP/s/W for case-study 1 and case-study 2, respectively, which is comparable to the state-of-the-art results. Finally, we compared our tiny FPGA hardware implementation results with off-the-shelf resource-constrained devices and showed our implementation is faster and consumed less power compared to the off-the-shelf resource-constrained devices.

Iterative method for large-scale Timoshenko beam models assessed on commercial-grade paperboard

AbstractLarge-scale structural simulations based on micro-mechanical models of paper products require extensive numerical resources and time. In such models, the fibrous material is often represented by connected beams. Whereas previous micro-mechanical simulations have been restricted to smaller sample problems, large-scale micro-mechanical models are considered here. These large-scale simulations are possible on a non-specialized desktop computer with 128GB of RAM using an iterative method developed for network models and based on domain decomposition. Moreover, this method is parallelizable and is also well-suited for computational clusters. In this work, the proposed memory-efficient iterative method is numerically validated for linear systems resulting from large networks of Timoshenko beams. Tensile stiffness and out-of-plane bending stiffness are simulated and validated for various commercial-grade three-ply paperboards consisting of layers composed of two different types of paper fibers. The results of these simulations show that a linear network model produces results consistent with theory and published experimental data

Hot rolled steel surface defect detection and classification using an automatic ensemble approach

Abstract This study introduces an ensemble-based Deep Neural Network (DNN) model for detecting defects on steel surfaces. The method suggested in this study classifies steel surface conditions into six possible fault categories, namely, crazing, inclusion, rolled in, pitted surface, scratches, and patches. The images undergo preprocessing and extraction of features in spatial and frequency domains using image segmentation techniques such as grey level difference method (GLDM), fast Fourier Transform (FFT), grey level co-occurrence matrix (GLCM), texture analysis and discrete wavelet transform (DWT). The ensembling of image features into a fused feature pool is carried out after the preprocessing of input images that are provided as input to a light-weight neural network model for training and testing. The performance of the model is comprehensively evaluated via an ablation study both before and after ensembling. In addition, the model capability is effectively analyzed using receiver operating characteristics (ROC) curve, confusion matrix from which classification accuracy of the model could be obtained and other parameters including precision and f1-score. It was observed that the proposed deep learning network presents phenomenally high accuracy of 99.72% for detection and classification of steel surface faults. This result was found to be superior when compared with the performance of the same neural network over each feature type individually. This study also compares the classification results of the model built based on the ensembled feature set with the results of various other classification approaches available in literature. The ensemble-based model could potentially be integrated into existing inspection systems for real-time, efficient and robust condition monitoring of steel surfaces.

Neural networks-based spring element for second-order analysis of pile-supported structures with nonlinear soil- structure interaction

This paper proposes a novel structural analysis approach, the neural networks-based spring element (NNSE) method, to synergize machine learning (ML) techniques with the line finite element method (LFEM) for the second-order analysis method of pile-supported structures. Traditional LFEM, widely used in upper structure design, showcases limitations in efficiently modeling complex Soil-Structure Interaction (SSI) along piles since it requires dense element mesh for accuracy. Conversely, ML offers a mesh-free alternative for analyzing single piles but struggles in simulating pile-supported structures as the training sample collection for large-scale problems might be unbearable. This paper addresses these challenges by proposing a new analysis framework to utilize the neural network (NN) model, which only describes structural responses of single piles, for the simulation of entire pile-supported structures. In the proposed method, the NN model is not directly used for structural analysis but employed to formulate a new spring element named the NNSE to model single piles in pile-supported structures. This NNSE can be seamlessly implemented within the existing LFEM framework to analyze pile-supported structures, eliminating the dense mesh requirement for single piles and thereby significantly improving the computational efficiency. Extensive examples are provided to verify the effectiveness of the proposed method, indicating its potential in promoting the second-order analysis method to the design of pile-supported structures.

Gait Impairment Analysis Using Silhouette Sinogram Signals and Assisted Knowledge Learning

The analysis of body motion is a valuable tool in the assessment and diagnosis of gait impairments, particularly those related to neurological disorders. In this study, we propose a novel automated system leveraging artificial intelligence for efficiently analyzing gait impairment from video-recorded images. The proposed methodology encompasses three key aspects. First, we generate a novel one-dimensional representation of each silhouette image, termed a silhouette sinogram, by computing the distance and angle between the centroid and each detected boundary points. This process enables us to effectively utilize relative variations in motion at different angles to detect gait patterns. Second, a one-dimensional convolutional neural network (1D CNN) model is developed and trained by incorporating the consecutive silhouette sinogram signals of silhouette frames to capture spatiotemporal information via assisted knowledge learning. This process allows the network to capture a broader context and temporal dependencies within the gait cycle, enabling a more accurate diagnosis of gait abnormalities. This study conducts training and an evaluation utilizing the publicly accessible INIT GAIT database. Finally, two evaluation schemes are employed: one leveraging individual silhouette frames and the other operating at the subject level, utilizing a majority voting technique. The outcomes of the proposed method showed superior enhancements in gait impairment recognition, with overall F1-scores of 100%, 90.62%, and 77.32% when evaluated based on sinogram signals, and 100%, 100%, and 83.33% when evaluated based on the subject level, for cases involving two, four, and six gait abnormalities, respectively. In conclusion, by comparing the observed locomotor function to a conventional gait pattern often seen in healthy individuals, the recommended approach allows for a quantitative and non-invasive evaluation of locomotion.

Deep learning algorithm development for early detection of Botrytis cinerea infected strawberry fruit using hyperspectral fluorescence imaging

Botrytis cinerea is a strawberry disease that causes economic loss worldwide. If a disease outbreak occurs during storage or transportation, it can spread rapidly to neighboring objects; thus, there is a need to develop early diagnostic techniques to prevent it. In this study, we developed a method to rapidly and nondestructively determine the infection stage in strawberry fruit using hyperspectral fluorescence imaging. ‘Keumsil’ cultivar strawberries were used, and hyperspectral fluorescence images were acquired over 144 h in control and inoculation groups. Strawberries were categorized into four infection stages based on visible mold spores: healthy, asymptomatic, infected, and after-infected. Hyperspectral fluorescence spectra were extracted to develop a one-dimensional convolutional neural network (1D-CNN) model based on partial least squares-discriminant analysis (PLS-DA), VGG-19, and ResNet-50; data augmentation techniques and six spectral preprocessing techniques were applied to the datasets. The application of data augmentation techniques improved the performances of the PLS-DA and 1D-CNN models in determining the infection stage. The performance of the ResNet-50-based 1D-CNN model with mean normalization data and data augmentation technique was the best, with 96.88% precision, 96.87% recall, 96.85% F1-score, and 96.86% accuracy. The results of this study showed that it is possible to determine the infection stage of Botrytis cinerea on strawberry fruit using hyperspectral fluorescence imaging and 1D-CNN techniques. This technology is expected to be applied for the early detection of Botrytis cinerea in strawberry growth, postharvest sorting and packing, and distribution stages.

Enhanced Solar Coronal Imaging: A GAN Approach with Fused Attention and Perceptual Quality Enhancement

The activity of the solar corona has a significant impact on all aspects of human life. People typically use images obtained from astronomical telescopes to observe coronal activities, among which the Atmospheric Imaging Assembly (AIA) of the Solar Dynamics Observatory (SDO) is particularly widely used. However, due to resolution limitations, we have begun to study the application of generative adversarial network super-resolution techniques to enhance the image data quality for a clearer observation of the fine structures and dynamic processes in the solar atmosphere, which improves the prediction accuracy of solar activities. We aligned SDO/AIA images with images from the High-Resolution Coronal Imager (Hi-C) to create a dataset. This research proposes a new super-resolution method named SAFCSRGAN, which includes a spatial attention module that incorporates channel information, allowing the network model to better capture the corona’s features. A Charbonnier loss function was introduced to enhance the perceptual quality of the super-resolution images. Compared to the original method using ESRGAN, our method achieved an 11.9% increase in Peak Signal-to-Noise Ratio (PSNR) and a 4.8% increase in Structural Similarity (SSIM). Additionally, we introduced two perceptual image quality assessment metrics, the Natural Image Quality Evaluator (NIQE) and Learned Perceptual Image Patch Similarity (LPIPS), which improved perceptual quality by 10.8% and 1.3%, respectively. Finally, our experiments demonstrated that our improved model surpasses other models in restoring the details of coronal images.

PointGAT: A Quantum Chemical Property Prediction Model Integrating Graph Attention and 3D Geometry.

Predicting quantum chemical properties is a fundamental challenge for computational chemistry. While the development of graph neural networks has advanced molecular representation learning and property prediction, their performance could be further enhanced by incorporating three-dimensional (3D) structural geometry into two-dimensional (2D) molecular graph representation. In this study, we introduce the PointGAT model for quantum molecular property prediction, which integrates 3D molecular coordinates with graph-attention modeling. Comparison with other current models in molecular prediction tasks showed that PointGAT could provide higher predictive accuracy in various benchmark data sets from MoleculeNet, including ESOL, FreeSolv, Lipop, HIV, and 6 out of 12 tasks of the QM9 data set. To further examine PointGAT prediction of quantum mechanical (QM) energies, we constructed a C10 data set comprising 11,841 charged and chiral carbocation intermediates with QM energies calculated at the DM21/6-31G*//B3LYP/6-31G* levels. Notably, PointGAT achieved an R2 value of 0.950 and an MAE of 1.616 kcal/mol, outperforming even the best-performing graph neural network model with a reduction of 0.216 kcal/mol in MAE and an improvement of 0.050 in R2. Additional ablation studies indicated that incorporating molecular geometry into the model resulted in markedly higher predictive accuracy, reducing the MAE value from 1.802 to 1.616 kcal/mol. Moreover, visualization of PointGAT atomic attention weights suggested its predictions were interpretable. Findings in this study support the application of PointGAT as a powerful and versatile tool for quantum chemical property prediction that can facilitate high-accuracy modeling for fundamental exploration of chemical space as well as drug design and molecular engineering.

An artificial neural network model for determining stress concentration factors for fatigue design of tubular T-joint under compressive loads

PurposeThe stress concentration factor (SCF) is commonly utilized to assess the fatigue life of a tubular T-joint in offshore structures. Parametric equations derived from experimental testing and finite element analysis (FEA) are utilized to estimate the SCF efficiently. The mathematical equations provide the SCF at the crown and saddle of tubular T-joints for various load scenarios. Offshore structures are subjected to a wide range of stresses from all directions, and the hotspot stress might occur anywhere along the brace. It is critical to incorporate stress distribution since using the single-point SCF equation can lead to inaccurate hotspot stress and fatigue life estimates. As far as we know, there are no equations available to determine the SCF around the axis of the brace.Design/methodology/approachA mathematical model based on the training weights and biases of artificial neural networks (ANNs) is presented to predict SCF. 625 FEA simulations were conducted to obtain SCF data to train the ANN.FindingsUsing real data, this ANN was used to create mathematical formulas for determining the SCF. The equations can calculate the SCF with a percentage error of less than 6%.Practical implicationsEngineers in practice can use the equations to compute the hotspot stress precisely and rapidly, thereby minimizing risks linked to fatigue failure of offshore structures and assuring their longevity and reliability. Our research contributes to enhancing the safety and reliability of offshore structures by facilitating more precise assessments of stress distribution.Originality/valuePrecisely determining the SCF for the fatigue life of offshore structures reduces the potential hazards associated with fatigue failure, thereby guaranteeing their longevity and reliability. The present study offers a systematic approach for using FEA and ANN to calculate the stress distribution along the weld toe and the SCF in T-joints since ANNs are better at approximating complex phenomena than standard data fitting techniques. Once a database of parametric equations is available, it can be used to rapidly approximate the SCF, unlike experimentation, which is costly and FEA, which is time consuming.

Environmental Risk Identification and Green Finance Development Based on Multi-scale Fusion Recognition Network

AbstractThis paper aims to enhance the resilience of financial enterprises against environmental risks by leveraging financial data analysis tools. The approach involves designing environmental risk assessment indicators and rating criteria. The study utilizes a convolutional neural network model extended by a multi-scale feature fusion module to analyze environmental risk information in the industry. The proposed model achieves impressive results with accuracy (Acc), precision (P), recall (R), and F1 scores reaching 99.09, 96.31, 95.32, and 95.64, respectively. These metrics outperform those of comparison models. The success of this model is anticipated to pave the way for the transformation of green finance through automated industry-level environmental risk assessment. Furthermore, the method’s adaptability extends beyond environmental risks, offering a scalable solution for identifying and assessing environmental risks in various contexts.

TEDNet: Twin Encoder Decoder Neural Network for 2D Camera and LiDAR Road Detection

Abstract Robust road surface estimation is required for autonomous ground vehicles to navigate safely. Despite it becoming one of the main targets for autonomous mobility researchers in recent years, it is still an open problem in which cameras and LiDAR sensors have demonstrated to be adequate to predict the position, size and shape of the road a vehicle is driving on in different environments. In this work, a novel Convolutional Neural Network model is proposed for the accurate estimation of the roadway surface. Furthermore, an ablation study has been conducted to investigate how different encoding strategies affect model performance, testing 6 slightly different neural network architectures. Our model is based on the use of a Twin Encoder–Decoder Neural Network (TEDNet) for independent camera and LiDAR feature extraction and has been trained and evaluated on the Kitti–Road dataset. Bird’s Eye View projections of the camera and LiDAR data are used in this model to perform semantic segmentation on whether each pixel belongs to the road surface. The proposed method performs among other state-of-the-art methods and operates at the same frame rate as the LiDAR and cameras, so it is adequate for its use in real-time applications.

Impact of Real-Time Blind Slat Angle Control on Reducing Total Energy Consumption of a Library Building Using an Optimized Artificial Neural Network Model: A Case Study of the Six Moroccan Thermal Zones

Abstract The prevalence of extensive glazed areas in contemporary buildings contributes significantly to solar radiation infiltration, elevating energy demands and causing discomfort for occupants. Window shading devices play a pivotal role in addressing this challenge. This paper presents the development and optimization of an artificial neural network (ANN) predictive model, designed to enable real-time control of slat angles by predicting total energy loads, specifically during summer (for cooling and lighting purposes). The refined model demonstrates high precision, achieving a normalized root mean square error (nRMSE) of approximately 1.72% and a correlation coefficient (R) of around 0.999, despite utilizing limited meteorological data. Key inputs for the model include solar radiation, solar altitude, and external temperature, with a particular focus on slat reflectivity. The study assesses the efficiency of three slat types based on their reflectivity: high (80%), medium (50%), and low (20%). Additionally, the research explores the impact of window-to-wall ratio (WWR) values on the control system's efficacy, revealing a positive correlation between higher WWR values and improved energy savings through ANN slat angle control. Furthermore, the study extends the applicability of the ANN model to the six thermal zones in Morocco, affirming its generalization across diverse environmental conditions.

Online identification and feed-forward compensation of nonlinear friction in servo system based on RBF neural network model

In this paper, an online identification and compensation method of nonlinear friction based on radial basis function (RBF) neural network model is proposed for the influence of nonlinear friction on machining accuracy in the low speed process of servo feed system of CNC machine tools. First, a three-layer single-input-output RBF neural network model is established for describing the nonlinear friction of servo feeding system. Second, the neural network online learning algorithm is improved based on adaptive gain, which improves the stability and accuracy of the algorithm. Finally, experiments were carried out on a three-axis milling machine to compensate the friction in the servo feed system in real time based on the online identification results. The results show that the method can effectively improve the online identification accuracy and convergence rate, and effectively improved the low-speed performance of the servo feed system.

Interpretable baseflow segmentation and prediction based on numerical experiments and deep learning

Baseflow is a crucial water source in the inland river basins of high-cold mountainous region, playing a significant role in maintaining runoff stability. It is challenging to select the most suitable baseflow separation method in data-scarce high-cold mountainous region and to evaluate effects of climate factors and underlying surface changes on baseflow variability and seasonal distribution characteristics. Here we attempt to address how meteorological factors and underlying surface changes affect baseflow using the Grey Wolf Optimizer Digital Filter Method (GWO-DFM) for rapid baseflow separation and the Long Short-Term Memory (LSTM) neural network model for baseflow prediction, clarifying interpretability of the LSTM model in baseflow forecasting. The proposed method was successfully implemented using a 63-year time series (1958–2020) of flow data from the Tai Lan River (TLR) basin in the high-cold mountainous region, along with 21 years of ERA5-land meteorological data and MODIS data (2000–2020). The results indicate that: (1) GWO-DFM can rapidly identify the optimal filtering parameters. It employs the arithmetic average of three methods, namely Chapman, Chapman-Maxwell and Eckhardt filter, as the best baseflow separation approach for the TLR basin. Additionally, the baseflow significantly increases after the second mutation of the baseflow rate. (2) Baseflow sources are mainly influenced by precipitation infiltration, glacier frozen soil layers, and seasonal ponding. (3) Solar radiation, temperature, precipitation, and NDVI are the primary factors influencing baseflow changes, with Nash-Sutcliffe efficiency coefficients exceeding 0.78 in both the LSTM model training and prediction periods. (4) Changes in baseflow are most influenced by solar radiation, temperature, and NDVI. This study systematically analyzes the changes in baseflow and response mechanisms in high-cold mountainous region, contributing to the management of water resources in mountainous basins under changing environmental conditions.

Quality Analysis and Shelf-Life Prediction of Antarctic Krill (Euphausia superba) Sauce Based on Kinetic Model and Back Propagation Neural Network Model

The study is aimed at determining how the quality of Antarctic krill (Euphausia superba) sauce (AkS) changed over time, including changes in color, moisture content, acid value (AV), peroxide value (POV), thiobarbituric acid reactive substance (TBARS), aerobic plate count, and sensory score. Quality variations and shelf life of AkS were estimated using kinetic model and back propagation (BP) neural network model. The results showed that sensory score, moisture content, and a∗ values of AkS declined as storage temperature increased at 4, 25, and 37°C. In addition, the L∗ values, b∗ values, AV, POV, and TBARS of AkS increased as storage duration increased, indicating that high storage temperature of the samples accelerated quality degradation. The primary reason for AkS degradation was the oxidation of proteins and lipids. The POV, TBARS, and total sensory evaluation rating exhibited a highly significant correlation, and therefore, POV and TBARS were selected as the indicators for the two models. The BP neural network outperformed the kinetic model in predicting quality changes over the whole storage period, with relative errors of less than 10%. In terms of shelf-life prediction, the BP neural network’s relative errors were 11.76% and 13.39% in POV and TBARS, respectively. POV and TBARS had experimental shelf lengths of 119 and 142 d, respectively. Compared with the kinetic model, the BP neural network model predicted the quality changes and shelf life of AkS with greater accuracy and stability. The findings offer fundamental insights and innovative concepts for the production of high-value Antarctic krill products, as well as the exploitation of Antarctic krill resources.

The reliability analysis and experiment verification of pressure spherical model for deep sea submersible based on data BP and machine learning technology

Spherical pressure-resistant shells, as a universal structural component of deep-sea submersibles, provide a safe and normal operating environment for personnel and internal equipment. In the paper it presented and optimized the BP neural network model based on a genetic algorithm (GA) accordingly, and the method and accuracy are validated through by a beam model. Simultaneously focusing on steel spherical shells, the study proposed a dataset that captures the influence of the primary dimension of the shell (radius-to-thickness ratio, R/t) on the critical pressure response. The genetic algorithm is employed to optimize the back propagation (BP) neural network model for predicting critical pressure. The structural reliability is adopted as a design criterion to determinate and optimize the geometric parameters and critical pressure of the spherical shell structure. Finally, an ultra-high-strength steel spherical model is designed, constructed and meanwhile collapse pressure tests are accomplished to verify the accuracy of the presented improved BP neural network model based on the computational reliability method. The results reveal that the machine learning optimization design method proposed in this paper can effectively enhance the accuracy of critical pressure predictions and the precision of reliability assessments for deep-sea spherical shells.

An efficient deep neural network model for tuberculosis detection using chest X-ray images

An efficient deep neural network model for tuberculosis detection using chest X-ray images