Neural Network Model Research Articles

Tripartite interaction representation algorithm for crystal graph neural networks

Driven by emerging research paradigms, the application of artificial intelligence models presents innovative tools for designing materials and optimizing their performance. In the field of materials science, there is a current research emphasis on exploring techniques for characterizing material structures to achieve precise descriptions. This paper proposes a crystal graph convolution neural network model that incorporates a tripartite interaction approach. The model not only incorporates atomic information, bond lengths, and bond angles but also offers a method for updating atoms and bond lengths, facilitating accurate descriptions of crystal structures by capturing implicit structural information. Focusing on predicting the formation energy of crystalline compounds, our results demonstrate improved predictive accuracy compared to existing representation algorithms. The average error of the formation energy in the random dataset, demonstrating robust generalization, is merely 0.048eV/atom, with an impressive R2\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$R^{2}$$\\end{document} value of 0.994. Additionally, this paper establishes a crystal graph neural network framework for predicting algorithm performance. By integrating automatic parallel algorithms and an automated process, we achieve a synthesis of these techniques, enhancing computational efficiency and streamlining algorithm usage.

Optimising Neural Networks for Enhanced Fracture Density Prediction in Surrounding Rock of Coalbed Methane Reservoir

ABSTRACTFractures influence the mechanical strength of coal roof and floor, constraining the design of hydraulic fracturing for coalbed methane production. Currently, the predominant approach involves the integration of petrophysical logging with machine learning for fracture prediction. Nevertheless, challenges exist regarding the model's accuracy. In this study, we present a novel approach to predict fracture density. Our method optimises a back‐propagation (BP) neural network and utilises principal component analysis for feature extraction. We employ logging parameters (density, compensated neutron and acoustic time difference) obtained from Shouyang Block well SY‐1 and fracture density data from electrical imaging logging to construct the FVDC model's dataset. The BP neural network model is optimised using the Sparrow Search algorithm and Tent Chaotic Mapping. The results demonstrate a substantial enhancement over the BP neural network model, with reductions of 80.102% in mean absolute error, 94.182% in mean square error, 75.879% in root mean square error and 79.764% in mean absolute percentage error. When considering accuracy, the optimised model (97.098%) surpasses the support vector regression model (96.478%), the random forest model (94.404%) and the BP neural network model (85.657%). Scalability testing for the optimised model was conducted using data from well SY‐2, yielding a remarkable prediction accuracy of 96.775%. This performance exceeds that of the BP neural network (with an accuracy of 85.102%), as well as the random forest and support vector regression models (with accuracies of 91.234% and 90.384%, respectively). These results underscore the potential of well logging and machine learning in FVDC prediction.

Calibration of discrete meta-parameters of bamboo flour based on magnitude analysis and BP neural network.

In the research and development of technology and equipment for bamboo products deep processing, such as filling, drying, and medicinal use of bamboo flour (BF), the poor compaction and fluidity of BF materials entails the need for accurate discrete element model (DEM) and BF parameters to provide a reference for the simulation of BF processing operationsand the development of related equipment. The average particle size of the 5 types of BFs ranges from 0.136 mm to 0.293 mm, and the small particle size of BF particles causes to the number of BF particles in bamboo processing equipment to reach tens of millions or even billions. When conventional methods are used for simulation, ordinary computers cannot provide the required computing power. To address the aforementioned challenges, this paper proposes a calibration method for the discrete element contact parameters of BFs based on dimensional analysis and a back propagation (BP) neural network. Using particle scaling theory and dimensional analysis methods, the average particle size of the BF was increased to 1 mm, and the main discrete element contact parameters of the five types of BF to be tested were used as input layers. The injection method and sidewall collapse method were used to obtain the angle of repose (AR) as the output layer. Fifty groups were randomly selected using MATLAB for EDEM simulation, and the simulation results were trained using the BP neural network algorithm; an ideal neural network model was obtained, the discrete element parameters of different BFs were predicted, and physical experiments were performed to verify two types of AR and mold hole compression under calibrated parameters. The relative error between the simulated AR obtained through calibration parameters and the physical experimental values is less than 2.3%. Through BF parameter validity verification, the simulated maximum compression displacement and compression ratio after stabilization were 34.81 mm and 0.477, which were close to the actual experimental results of 34.77 mm and 0.461, respectively, verifying the accuracy of the neural network prediction model. The research results provide a reference for the simulation of BF processing operations and the development of related equipment.

Sub-millisecond keyhole pore detection in laser powder bed fusion using sound and light sensors and machine learning

Abstract Laser powder bed fusion is a mainstream additive manufacturing technology widely used to manufacture complex parts in prominent sectors, including aerospace, biomedical, and automotive industries. However, during the printing process, the presence of an unstable vapor depression can lead to a type of defect called keyhole porosity, which is detrimental to the part quality. In this study, we developed an effective approach to locally detect the generation of keyhole pores during the printing process by leveraging machine learning and a suite of optical and acoustic sensors. Simultaneous synchrotron x-ray imaging allows the direct visualization of pore generation events inside the sample, offering high-fidelity ground truth. A neural network model adopting SqueezeNet architecture using single-sensor data was developed to evaluate the fidelity of each sensor for capturing keyhole pore generation events. Our comparative study shows that the near infrared images gave the highest prediction accuracy, followed by 100kHz and 20kHz microphones, and the photodiode sensitive to processing laser wavelength had the lowest accuracy. Using a single sensor, over 90% prediction accuracy can be achieved with a temporal resolution as short as 0.1 ms. A data fusion scheme was also developed with features extracted using SqueezeNet neural network architecture and classification using different machine learning algorithms. Our work demonstrates the correlation between the characteristic optical and acoustic emissions and the keyhole oscillation behavior, and thereby provides strong physics support for the machine learning approach.

Response of soybean Cd to soil Cd and pH and its associated health risk in a high geological background area in Guizhou Province, Southwest China.

This study comprehensively examined the accumulation of cadmium (Cd) in soybeans grown in low- and high-Cd soils around the high geological background areas in Guizhou province. The aim was to analyze the relationship between soybean Cd and soil pH and soil Cd, alongside assessing the potential carcinogenic and non-carcinogenic risks associated with Cd in soybeans. Cd content of soybeans cultivated in the high-Cd area (0.430 mg/kg) was significantly higher than that in low-Cd areas (0.156 mg/kg) (P < 0.05). Biological concentration factors (BCFs) of soybean for Cd in low- and high-Cd areas were 0.282 and 0.314, respectively, with no significant differences (P > 0.05). Multiple linear regression results indicated that soil pH was a determining factor for Cd accumulation in soybeans in both areas. Furthermore, soil pH and soil Cd could accurately predict Cd accumulation in soybeans according to the neural network model. These findings suggest that regulating soil pH could reduce Cd accumulation in soybeans in areas with high geological background. In both areas, there was no significant non-carcinogenic risk for the adult population (HQ value < 1) through soybean consumption. However, according to the Monte Carlo model, the percentage of Cd in soybeans exceeding the acceptable range (CR value > 1.00 × 10 -04) in areas was 99.18%, indicating an unacceptable carcinogenic risk for the adult population. Our discussion revealed that reducing the soybean intake and increasing soil pH did not effectively lower the carcinogenic risk of Cd in soybeans to an acceptable range (CR value ≤ 1.00 × 10 -04). These findings necessitate further exploration of alternative remediation strategies to ensure the safe production of soybeans, such as screening for low-Cd accumulation soybean varieties and implementing the combined remediation strategies.

Forecasting Orange Juice Futures: LSTM, ConvLSTM, and Traditional Models Across Trading Horizons

This study evaluated the forecasting accuracy of various models over 5-day and 10-day trading horizons to predict the prices of orange juice futures (OJ = F). The analysis included traditional models like Autoregressive Integrated Moving Average (ARIMA) and advanced neural network models such as Long Short-Term Memory (LSTM), Recurrent Neural Network (RNN), Backpropagation Neural Network (BPNN), Support Vector Regression (SVR), and Convolutional Long Short-Term Memory (ConvLSTM), incorporating factors like the Commodities Index and the S&amp;P500 Index. We employed loss function metrics and various tests to assess model performance. The results indicated that for the 5-day horizon, the LSTM and ConvLSTM consistently outperformed the other models. LSTM achieved the lowest error rates and demonstrated superior capability in capturing temporal dependencies, especially in single-factor and S&amp;P500 Index predictions. ConvLSTM also performed strongly, effectively modeling spatial and temporal data patterns. In the 10-day horizon, similar trends were observed. LSTM and ConvLSTM models had significantly lower errors and better alignment with actual values. The BPNN model performed well when all factors were included, and the SVR model maintained consistent accuracy, particularly for single-factor predictions. The Diebold–Mariano (DM) test indicated significant differences in forecasting accuracy, favoring advanced neural network models. In addition, incorporating multiple influencing factors further improved predictive performance, enhancing investment outcomes and reducing risk.

Automated volumetric analysis of the inner ear fluid space from hydrops magnetic resonance imaging using 3D neural networks

Due to the development of magnetic resonance (MR) imaging processing technology, image-based identification of endolymphatic hydrops (EH) has played an important role in understanding inner ear illnesses, such as Meniere’s disease or fluctuating sensorineural hearing loss. We segmented the inner ear, consisting of the cochlea, vestibule, and semicircular canals, using a 3D-based deep neural network model for accurate and automated EH volume ratio calculations. We built a dataset of MR cisternography (MRC) and HYDROPS-Mi2 stacks labeled with the segmentation of the perilymph fluid space and endolymph fluid space of the inner ear to devise a 3D segmentation deep neural network model. End-to-end learning was used to segment the perilymph fluid and the endolymph fluid spaces simultaneously using aligned pair data of the MRC and HYDROPS-Mi2 stacks. Consequently, the segmentation performance of the total fluid space and endolymph fluid space had Dice similarity coefficients of 0.9574 and 0.9186, respectively. In addition, the EH volume ratio calculated by experienced otologists and the EH volume ratio value predicted by the proposed deep learning model showed high agreement according to the interclass correlation coefficient (ICC) and Bland–Altman plot analysis.

Deep Learning-Based Stock Price Prediction Using LSTM Model

The stock market is a vital component of the broader financial system, with its dynamics closely linked to economic growth. The challenges associated with analyzing and forecasting stock prices have persisted since the inception of financial markets. By examining historical transaction data, latent opportunities for profit can be uncovered, providing valuable insights for both institutional and individual investors to make more informed decisions. This study focuses on analyzing historical transaction data from four banks to predict closing price trends. Various models, including decision trees, random forests, and Long Short-Term Memory (LSTM) networks, are employed to forecast stock price movements. Historical stock transaction data serves as the input for training these models, which are then used to predict upward or downward stock price trends. The study’s empirical results indicate that these methods are effective to a degree in predicting stock price movements. The LSTM-based deep neural network model, in particular, demonstrates a commendable level of predictive accuracy. This conclusion is reached following a thorough evaluation of model performance, highlighting the potential of LSTM models in stock market forecasting. The findings offer significant implications for advancing financial forecasting approaches, thereby improving the decision-making capabilities of investors and financial institutions.

Deep Learning‐Driven Modeling of Dynamic Acoustic Sensing in Biomimetic Soft‐Robotic Pinnae

ABSTRACTBiological function often depends on complex mechanisms of a dynamic, time‐variant nature. An example is certain bat species (horseshoe bats—Rhinolophidae) that use intricate pinna musculatures to execute a variety of pinna deformations. While prior work has indicated the potential significance of these motions for sensory information encoding, it remains unclear how the complex time‐variant pinna geometries could be controlled to enhance sensory performance. To address this issue, this work has investigated deep neural network models as digital twins for biomimetic pinnae. The networks were trained to predict the acoustic impacts of the deformed pinna geometries. A total of three network architectures have been evaluated for this purpose using physical numerical simulations (boundary element method) as ground truth. The networks predicted the acoustic beampattern function from pinna shape or even directly from the states of actuators that were used to deform the pinna shapes in simulation. Inserting prior knowledge in the form of beam‐shaped basis functions did not improve network performance. The ability of the networks to produce beampattern predictions with low computational effort (in about three milliseconds each) should lend itself readily to supporting learning methods such as deep reinforcement learning that require many such functional evaluations.

Pushing the boundaries for fuel discovery with a multiview features fusion approach

AbstractGlobal warming poses a serious challenge to the human environment, prompting us to rapidly develop new environmentally friendly fuels. However, the time and cost required to determine the physical properties of fuels are constrained by the related industries. In this paper, we propose a multiview features fusion method based on neural networks. This method uses the eight graph neural networks models as the basis of the multichannel network coordinator to extract the compound's molecular feature; also the functional groups in the compound are embedded with molecule weight as functional groups feature, and finally, combining the molecular view with the functional groups view for the prediction of compound flash point (FP). We used a data set consisting of 2200 hydrocarbons and oxygenated compounds for model training and testing. The results show that the model performance is improved in both after feature fusion. Finally, the ablation experiments demonstrate that the use of this method is effective and provides a new idea for fast and accurate screening of fuels. The Attentive FP‐FG model was the most effective, with a mean absolute error of 4.395 K, root mean square error of 5.854 K, and R‐squared score (R2) of 0.986.

A self-organizing map clustering adaptive artificial neural network model for analysis critical success factors in different phases of MTO projects case study: electrical equipment manufacturer and supplier in Iran

PurposeThe purpose of this paper is to investigate the benefit of critical success factors (CSFs) clustering in different phases of make-to-order (MTO) projects and develop standards for management.Design/methodology/approachThis study is based on a questionnaire survey. First of all, collecting data by structured interviews, relying on a questionnaire and second from leader contractors who are active in the engineering and steel industry (in Iran). So, the requirements and objective of the research are presented to the top management of MTO projects to gain their support in data collection. Then 20 CSFs were identified by the literature review so a questionnaire survey was prepared for the CSFs assessment and interview with the experts. Finally analyzing the importance and performance of CSFs in project phases and cluster them in different project phases with self-organizing map as one of the artificial neural network (ANN) approaches due to high predictive accuracy. Review the research result with the top management of MTO project and examine the results obtained from neural networks and validation indices.FindingsCluster analysis shows that the implementation phase is the most important stage in MTO organizations and the other phases like feasibility and start-up, design and planning, delivery and end-phase should be also considered as effective phases in determining the level of organization performance. Different industries with additional data at different periodic times will verify the result. Furthermore, testing the other ANN model will improve risk analysis and could shift this classification approach to a regression type.Research limitations/implicationsThe main limitation of the research is related to the sample. Research findings are limited to the time of data collection so validity is limited to the mentioned time. Different industries with additional data will verify the result. Furthermore, testing different ANN models such as K-MEANS, non-negative matrix factorization (NMF) analyses will improve risk analysis and could meet different classification results to find gaps.Practical implicationsIn this paper, CSF and project phase dimensions are viewed together which is necessary to meet better results for simplifying social and economic benefits. Merge the new findings and latest technologies could prepare the best results and enable managers to create a better framework or implement key factors for minimizing waste.Originality/valueThis paper moves the definition of MTO organizations beyond measuring cost, complexity and financial variables by clustering CSFs in different phases of projects. So, the results enable managers to use this concept in their daily production to minimize waste and could be implemented to efficiently choose factors.

Advanced neural network-based model for predicting court decisions on child custody

Predicting court rulings has gained attention over the past years. The court rulings are among the most important documents in all legal systems, profoundly impacting the lives of the children in case of divorce or separation. It is evident from literature that Natural language processing (NLP) and machine learning (ML) are widely used in the prediction of court rulings. In general, the court decisions comprise several pages and require a lot of space. In addition, extracting valuable information and predicting legal decisions task is difficult. Moreover, the legal system’s complexity and massive litigation make this problem more serious. Thus to solve this issue, we propose a new neural network-based model for predicting court decisions on child custody. Our proposed model efficiently performs an efficient search from a massive court decisions database and accurately identifies specific ones that especially deal with copyright claims. More specially, our proposed model performs a careful analysis of court decisions, especially on child custody, and pinpoints the plaintiff’s custody request, the court’s ruling, and the pivotal arguments. The working mechanism of our proposed model is performed in two phases. In the first phase, the isolation of pertinent sentences within the court ruling encapsulates the essence of the proceedings performed. In the second phase, these documents were annotated independently by using two legal professionals. In this phase, NLP and transformer-based models were employed and thus processed 3,000 annotated court rulings. We have used a massive dataset for the training and refining of our proposed model. The novelty of the proposed model is the integration of bidirectional encoder representations from transformers (BERT) and bidirectional long short-term memory (Bi_LSTM). The traditional methods are primarily based on support vector machines (SVM), and logistic regression. We have performed a comparison with the state-of-the-art model. The efficient results indicate that our proposed model efficiently navigates the complex terrain of legal language and court decision structures. The efficiency of the proposed model is measured in terms of the F1 score. The achieved results show that scores range from 0.66 to 0.93 and Kappa indices from 0.57 to 0.80 across the board. The performance is achieved at times surpassing the inter-annotator agreement, underscoring the model’s adeptness at extracting and understanding nuanced legal concepts. The efficient results proved the potential of the proposed neural network model, particularly those based on transformers, to effectively discern and categorize key elements within legal texts, even amidst the intricacies of judicial language and the layered complexity of appellate rulings.

Predicting chlorophyll-a dynamics in the Ashtamudi estuary using ANN and ANFIS techniques

Excessive nutrients and associated high chlorophyll-a concentration are of growing concern to the public health due to the occurrence of eutrophication in aquatic bodies. Continuous monitoring of chlorophyll-a has limitations, thereby necessitating the development of prediction models. Even though chlorophyll-a prediction models are numerous, they rarely fit into estuarine conditions, which is encountered with dynamic mixing of freshwater and sea water. Thus, the present study identifies the most appropriate driving factors that are relevant for estuarine conditions through chlorophyll-a prediction models developed using Artificial Neural Network and Adaptive Neuro Fuzzy Inference System (ANFIS) approaches in the Ashtamudi estuary, India. Salinity, which is an indirect measure of the estuarine mixing intensity, has been used to replicate the dynamic mixing in the estuary. The results indicate that the model incorporating salinity along with total nitrogen, total phosphorous and turbidity well-predicted chlorophyll-a concentration with a coefficient of determination ranging from 0.73 to 0.99. General Regression Neural Network (GRNN) and ANFIS models outperformed Back Propagation Neural Network (BPNN) and Recurrent Neural Network (RNN) models. Of the various ANFIS models, the ones that used Gaussian and Generalized Bell membership functions were most appropriate for the prediction of chlorophyll-a than the models with triangular and trapezoidal membership functions. The study identified that the models that considered salinity were less sensitive and can be used for modelling chlorophyll-a. The chlorophyll-a was observed to have a direct influence of salinity at salinity values greater than 5‰. The study highlighted that estuarine mixing further enhances the primary productivity through increased penetration of light leading to higher chlorophyll-a concentration. Further the study emplasized that the predictive models are important tools fitting well within estuarine environments due to its replicability.

LesionNet: an automated approach for skin lesion classification using SIFT features with customized convolutional neural network

Accurate detection of skin lesions through computer-aided diagnosis has emerged as a critical advancement in dermatology, addressing the inefficiencies and errors inherent in manual visual analysis. Despite the promise of automated diagnostic approaches, challenges such as image size variability, hair artifacts, color inconsistencies, ruler markers, low contrast, lesion dimension differences, and gel bubbles must be overcome. Researchers have made significant strides in binary classification problems, particularly in distinguishing melanocytic lesions from normal skin conditions. Leveraging the “MNIST HAM10000” dataset from the International Skin Image Collaboration, this study integrates Scale-Invariant Feature Transform (SIFT) features with a custom convolutional neural network model called LesionNet. The experimental results reveal the model's robustness, achieving an impressive accuracy of 99.28%. This high accuracy underscores the effectiveness of combining feature extraction techniques with advanced neural network models in enhancing the precision of skin lesion detection.

Research on Target Image Classification in Low-Light Night Vision.

In extremely dark conditions, low-light imaging may offer spectators a rich visual experience, which is important for both military and civic applications. However, the images taken in ultra-micro light environments usually have inherent defects such as extremely low brightness and contrast, a high noise level, and serious loss of scene details and colors, which leads to great challenges in the research of low-light image and object detection and classification. The low-light night vision image used as the study object in this work has an excessively dim overall picture and very little information about the screen's features. Three algorithms, HE, AHE, and CLAHE, were used to enhance and highlight the image. The effectiveness of these image enhancement methods is evaluated using metrics such as the peak signal-to-noise ratio and mean square error, and CLAHE was selected after comparison. The target image includes vehicles, people, license plates, and objects. The gray-level co-occurrence matrix (GLCM) was used to extract the texture features of the enhanced images, and the extracted image texture features were used as input to construct a backpropagation (BP) neural network classification model. Then, low-light image classification models were developed based on VGG16 and ResNet50 convolutional neural networks combined with low-light image enhancement algorithms. The experimental results show that the overall classification accuracy of the VGG16 convolutional neural network model is 92.1%. Compared with the BP and ResNet50 neural network models, the classification accuracy was increased by 4.5% and 2.3%, respectively, demonstrating its effectiveness in classifying low-light night vision targets.

LSTM-CRP: Algorithm-Hardware Co-Design and Implementation of Cache Replacement Policy Using Long Short-Term Memory

As deep learning has produced dramatic breakthroughs in many areas, it has motivated emerging studies on the combination between neural networks and cache replacement algorithms. However, deep learning is a poor fit for performing cache replacement in hardware implementation because its neural network models are impractically large and slow. Many studies have tried to use the guidance of the Belady algorithm to speed up the prediction of cache replacement. But it is still impractical to accurately predict the characteristics of future access addresses, introducing inaccuracy in the discrimination of complex access patterns. Therefore, this paper presents the LSTM-CRP algorithm as well as its efficient hardware implementation, which employs the long short-term memory (LSTM) for access pattern identification at run-time to guide cache replacement algorithm. LSTM-CRP first converts the address into a novel key according to the frequency of the access address and a virtual capacity of the cache, which has the advantages of low information redundancy and high timeliness. Using the key as the inputs of four offline-trained LSTM network-based predictors, LSTM-CRP can accurately classify different access patterns and identify current cache characteristics in a timely manner via an online set dueling mechanism on sampling caches. For efficient implementation, heterogeneous lightweight LSTM networks are dedicatedly constructed in LSTM-CRP to lower hardware overhead and inference delay. The experimental results show that LSTM-CRP was able to averagely improve the cache hit rate by 20.10%, 15.35%, 12.11% and 8.49% compared with LRU, RRIP, Hawkeye and Glider, respectively. Implemented on Xilinx XCVU9P FPGA at the cost of 15,973 LUTs and 1610 FF registers, LSTM-CRP was running at a 200 MHz frequency with 2.74 W power consumption.

TL-CStrans Net: a vision robot for table tennis player action recognition driven via CS-Transformer

Currently, the application of robotics technology in sports training and competitions is rapidly increasing. Traditional methods mainly rely on image or video data, neglecting the effective utilization of textual information. To address this issue, we propose: TL-CStrans Net: A vision robot for table tennis player action recognition driven via CS-Transformer. This is a multimodal approach that combines CS-Transformer, CLIP, and transfer learning techniques to effectively integrate visual and textual information. Firstly, we employ the CS-Transformer model as the neural computing backbone. By utilizing the CS-Transformer, we can effectively process visual information extracted from table tennis game scenes, enabling accurate stroke recognition. Then, we introduce the CLIP model, which combines computer vision and natural language processing. CLIP allows us to jointly learn representations of images and text, thereby aligning the visual and textual modalities. Finally, to reduce training and computational requirements, we leverage pre-trained CS-Transformer and CLIP models through transfer learning, which have already acquired knowledge from relevant domains, and apply them to table tennis stroke recognition tasks. Experimental results demonstrate the outstanding performance of TL-CStrans Net in table tennis stroke recognition. Our research is of significant importance in promoting the application of multimodal robotics technology in the field of sports and bridging the gap between neural computing, computer vision, and neuroscience.

Hand Gesture Recognition Using Ultrasonic Array with Machine Learning

In the field of gesture recognition technology, accurately detecting human gestures is crucial. In this research, ultrasonic transducers were utilized for gesture recognition. Due to the wide beamwidth of ultrasonic transducers, it is difficult to effectively distinguish between multiple objects within a single beam. However, they are effective at accurately identifying individual objects. To leverage this characteristic of the ultrasonic transducer as an advantage, this research involved constructing an ultrasonic array. This array was created by arranging eight transmitting transducers in a circular formation and placing a single receiving transducer at the center. Through this, a wide beam area was formed extensively, enabling the measurement of unrestricted movement of a single hand in the X, Y, and Z axes. Hand gesture data were collected at distances of 10 cm, 30 cm, 50 cm, 70 cm, and 90 cm from the array. The collected data were trained and tested using a customized Convolutional Neural Network (CNN) model, demonstrating high accuracy on raw data, which is most suitable for immediate interaction with computers. The proposed system achieved over 98% accuracy.

Capturing the attenuating influence of a Three Gorges Dam on downstream sediment loads using artificial neural networks

ABSTRACT River sediment dynamics have been significantly altered by dam constructions, with the Three Gorges Dam (TGD) on the Yangtze River being a prominent example. This study aims to quantitatively assess the attenuating impact of TGD operations on downstream sediment discharge using artificial neural network (ANN) models. Seven ANN models were developed for stations along the middle and lower Yangtze River, trained on measured water and sediment data. The models effectively captured the relationships between upstream inputs and downstream outputs. Results revealed that variations in sediment discharge at each station were predominantly influenced by changes in sediment input from its immediate upstream section, rather than by changes in water discharge. This suggests that the TGD's impact on downstream sediment transport was primarily due to its sediment trapping efficiency, rather than alterations in the hydrological regime. The influence diminished gradually farther downstream, attributed to tributary contributions and sediment transport dynamics. ANNs, adept at handling input uncertainties, underscore the importance of boundary conditions; the integration of data-driven and physics-based models illuminates sediment dynamics in regulated rivers. This approach provides insight into dam management implications and sediment transport processes.

Use of Deep-Learning-Accelerated Gradient Approximation for Reservoir Geological Parameter Estimation

The estimation of space-varying geological parameters is often not computationally affordable for high-dimensional subsurface reservoir modeling systems. The adjoint method is generally regarded as an efficient approach for obtaining analytical gradient and, thus, proceeding with the gradient-based iteration algorithm; however, the infeasible memory requirement and computational demands strictly prohibit its generic implementation, especially for high-dimensional problems. The autoregressive neural network (aNN) model, as a nonlinear surrogate approximation, has gradually received increasing popularity due to significant reduction of computational cost, but one prominent limitation is that the generic application of aNN to large-scale reservoir models inevitably poses challenges in the training procedure, which remains unresolved. To address this issue, model-order reduction could be a promising strategy, which enables us to train the neural network in a very efficient manner. A very popular projection-based linear reduction method, i.e., propel orthogonal decomposition (POD), is adopted to achieve dimensionality reduction. This paper presents an architecture of a projection-based autoregressive neural network that efficiently derives an easy-to-use adjoint model by the use of an auto-differentiation module inside the popular deep learning frameworks. This hybrid neural network proxy, referred to as POD-aNN, is capable of speeding up derivation of reduced-order adjoint models. The performance of POD-aNN is validated through a synthetic 2D subsurface transport model. The use of POD-aNN significantly reduces the computation cost while the accuracy remains. In addition, our proposed POD-aNN can easily obtain multiple posterior realizations for uncertainty evaluation. The developed POD-aNN emulator is a data-driven approach for reduced-order modeling of nonlinear dynamic systems and, thus, should be a very efficient modeling tool to address many engineering applications related to intensive simulation-based optimization.