Search

In the experiments of measuring the strength of materials under ramp compression, accurately determining in situ particle velocity is crucial for calculating material sound speed during loading–unloading path and materials strength under high pressure. This paper proposes a machine learning approach that utilizes Long Short-Term Memory (LSTM) neural networks and Bayesian optimization algorithms to enhance the analysis of data from ramp compression strength measurement experiments. This method leverages LSTM neural networks to uncover the complex relationship between the rear interface velocity of the sample and the in situ particle velocity in numerical simulations. By using a well-trained network model, it enables direct interpretation of experimental data, leading to accurate predictions of key physical quantities along the loading and unloading paths in ramp compression experiments. A comparative analysis between theoretical curves from numerical simulations and LSTM neural network predictions shows a high degree of consistency. This approach is applied to ramp compression experiments on Ta and CuCrZr materials, demonstrating superior accuracy over the free-surface approximation and incremental impedance matching methods. Additionally, this method relies solely on the equation of state during numerical computations, eliminating the need for the complex constitutive equations required by the transfer function method, thus enhancing data processing efficiency and practicality. [ABSTRACT FROM AUTHOR]

Microgrids are a development trend and have attracted a lot of attention worldwide. The control system plays a crucial role in implementing these systems and, due to their complexity, artificial intelligence techniques represent some enabling technologies for their future development and success. In this paper, we propose a novel formulation of an economic model predictive control (economic MPC) applied to a microgrid designed for a faculty building with the inclusion of a predictive model to deal with the energy demand disturbance using a recurrent neural network of the long short‐term memory (RNN‐LSTM). First, we develop a framework to identify an RNN‐LSTM using historical data registered by a smart three‐phase power quality analyzer to provide feedforward power demand predictions. Next, we present an economic MPC formulation that includes the prediction model for the disturbance within the optimization problem to be solved by the MPC strategy. We carried out simulations with different scenarios of energy consumption, available resources, and simulation times to highlight the results obtained and analyze the performance of the energy management system. In all cases, we observed the correct operation of the proposed control scheme, complying at all times with the objectives and operational restrictions imposed on the system. [ABSTRACT FROM AUTHOR]

Blending hydrogen into natural gas pipelines in utility tunnels is crucial for hydrogen applications, but introduces safety challenges. Therefore, this study proposes a new leakage detection method for hydrogen-blended natural gas (HBNG) pipelines in utility tunnels. This method employs computational fluid dynamics (CFD) simulations to generate a leakage diffusion dataset for HBNG pipelines. It then combines multi-task learning (MTL) with the long short-term memory (LSTM) network to develop an MTL-LSTM model for HBNG pipelines leakage detection in utility tunnels. The impact of hyperparameters on model accuracy is assessed using the dataset and the model is compared with other neural networks. Results indicate that after optimizing the hyperparameter combination, the MTL-LSTM model demonstrates higher accuracy and lower relative error in predicting the leakage location and rate of HBNG pipelines compared with other neural network models. This study provides technical support for the safe operation of HBNG pipelines in utility tunnels. • A leakage detection method for HBNG pipelines in utility tunnels is proposed. • The leakage diffusion dataset is constructed by CFD simulation. • The LSTM model based on multi-task learning is applied to pipeline leak detection. • The effectiveness of the proposed method is verified by model evaluation. [ABSTRACT FROM AUTHOR]

Bridge deflection serves as a vital and intuitive index for the evaluation of bridge safety. Temperature load has the greatest influence on the bridge deformation and studies on the temperature-induced deformation prediction of long-span bridge are in limited numbers. A digital prediction model based on deep learning in minute scale is established to study the bridge deflection caused by temperature. The wavelet transform (WT) is adopted to filter the high-frequency signals of the original deflection caused by the related load factors. Three different networks, long short-term memory (LSTM), bidirectional LSTM (Bi-LSTM), and Transformer variant, are studied and compared in the prediction process. Two different learning strategies considering different input data are also considered to optimize the prediction performance. The proposed prediction model is applied to the temperature induced deflection prediction of a multi-tower double-layer steel truss bridge. The results show that strategy A, which employs temperature time series data as input, is less effective than strategy B. Incorporating both temperature and deflection data as inputs is essential for predicting temperature-induced deflections. Moreover, the Transformer-variant network generally exhibits superior prediction performance compared to the LSTM and Bi-LSTM. The self-attention mechanism of the Transformer allows it to focus on key historical temperature points, thereby enhancing prediction accuracy. [ABSTRACT FROM AUTHOR]

To ensure structural safety and integrity, a novel framework is developed for detecting the loosening of multi-bolt connections using wavelet entropy of vibro-acoustic modulation (VAM) signals. Wavelet entropy is employed as the dynamic index to capture the intricate time-frequency characteristics that are indicative of the connection status. Taking the wavelet entropy vectors as input, the proposed framework distinguishes itself by integrating a Transformer model for high-dimensional feature extraction with the recursive feature elimination (RFE) for essential feature selection, followed by a support vector machine (SVM) model for classification. Specifically, the Transformer model with innovative positional encoding capability helps to extract the time-dependent transient features that are sensitive to the bolt loosening. The RFE process reduces the data dimensionality while discerning the diagnostic information for more accurate classification. Through the experiment on a four-bolt joint, the identification results with cross-validation showed high accuracy and robustness of the proposed framework across various loosening cases. It outperformed the traditional SVM, long short-term memory network (LSTM), convolutional neural network (CNN)-SVM models without and with RFE, as well as the Transformer-SVM model without RFE, achieving an accuracy increase of 15.72%, 11.74%, 9.47%, 5.49%, and 5.06%, respectively. The proposed framework was demonstrated to be able to learn the damage-sensitive features more effectively from wavelet entropy data, marking a significant advancement in the health monitoring of engineering structures with high-strength bolt connections. [ABSTRACT FROM AUTHOR]

Accurately predicting rainfall is still a difficult but crucial task in meteorological forecasting, with significant implications for agriculture, disaster planning, and water resource management. In the past, statistical methods were used to forecast rainfall, however these methods were less accurate due to data dynamics and other problems. To address the data issue in this study, Discrete Wavelet Transform (DWT) was employed, it improves rainfall data preprocessing by decomposing data, eliminating noise, risk of overfitting and concentrating on significant patterns. Overlapping Flexible Sliding Window (OFSW) technique is essential since it dynamically modifies window sizes according to data properties unlike the overlapping sliding windows. Leveraging the advancements in Deep Learning (DL), this paper presents rainfall prediction by employing DL algorithms Convolutional Neural Networks (CNNs), Long Short-Term Memory Networks (LSTMs), and Multilayer Perceptron (MLPs). Furthermore, the standalone models were fused to form a weighted average ensemble. The monthly data from Nigeria and the Kaduna region were gathered and used in forecasting. The Bi-LSTM model has shown better performance compared to other models in predicting rainfall univariate time series data. In this study, these models are integrated into an ensemble structure for prediction purposes. The obtained ensemble learning model harmonizes the advantages of CNNs, Bi-LSTMs, and MLPs. The results of the weighted average ensemble demonstrated the benefits of the ensemble model in rainfall prediction in Kaduna by achieving the least MSE of 0.0018,0.04242, 0.0303 of MSE, RMSE and MAE in the training phase respectively, while in the testing phase, it achieves, 0.0041, 0.0640 and 0.0447 of MSE, RMSE and MAE in the testing phase respectively. MSE, RMSE and MAE of 0.0017, 0.0412, and 0.0301 for Nigeria in the training phase respectively, while in the testing phase, 0.0042, 0.0648 and 0.0457 of MSE, RMSE and MAE were achieved. [ABSTRACT FROM AUTHOR]

Water resources are precious resources around the world, but pollution is becoming increasingly severe. While the degree of damage to water caused by organic pollutants is reflected by chemical oxygen demand, which is also an important indicator for hydrological monitoring. Therefore, accurate measurement of chemical oxygen demand is necessary. The rapid development of water quality detection by ultraviolet visible spectrum analysis as an efficient, convenient, and pollution-free new water quality detection method has been achieved. When we were processing ultraviolet visible spectral absorbance data, various environmental factors can affect the chemical oxygen demand light absorption value, thus affecting the accuracy of detection. So there are some goals, such as more comprehensively extracting the feature wavelength and reducing the interference to achieve better detection results. It is expected that the model can be applied to a portable water quality detection device. It is necessary to consider detection time, reduce model training parameters, and improve detection speed. Therefore, this article proposes Competitive Adaptive Weighted Sampling-Convolutional Neural Network-Long Short-Term Memory and Competitive Adaptive Weighted Sampling-Convolutional Long Short-Term Memory models for detecting chemical oxygen demand in water. To verify the effectiveness of the proposed model, the model was trained on the experimental dataset and the detection results of all models involved in the article were compared. It can be confirmed that the application of these two proposed algorithms results in less detection time and higher accuracy in concentration detection. [ABSTRACT FROM AUTHOR]

GNSS (Global Navigation Satellite System) time series are indispensable in geodesy, geophysics, and other Earth sciences, and serve as important tools for monitoring crustal deformation, plate tectonics, and other geodynamic phenomena. Analytical methods are used to improve the robustness and data quality of the results obtained from GNSS station position time series. The objective of this paper is to investigate the applicability of deep learning techniques in modeling and prediction studies on GNSS station position time series. The performance of 8 deep learning models, including Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Temporal Convolution Network (TCN), TCN-Leaky ReLU, TCN-Leaky ReLU-LSTM, Bidirectional LSTM, Bidirectional GRU, and Stack-LSTM, are analyzed based on the traditional Least Squares (LS) method and their ability to improve the prediction accuracy on three components of 9 GNSS stations in Western Turkey. The results show that the deep learning methods provide improvements in Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) by 45% and 53% in the East component, 44% and 51% in the North component, and 34% and 41% in the Up component, respectively, compared to the LS model. The study provides a detailed comparative analysis of model performance and demonstrates the performance of the Bi-LSTM and Bi-GRU and GRU models in handling high noise environments and complex transient changes at some stations. [ABSTRACT FROM AUTHOR]

Research on developing deep learning techniques for autonomous spacecraft relative navigation challenges is continuously growing in recent years. Adopting those techniques offers enhanced performance. However, such approaches also introduce heightened apprehensions regarding the trustability and security of such deep learning methods through their susceptibility to adversarial attacks. In this work, we propose a novel approach for adversarial attack detection for deep neural network-based relative pose estimation schemes based on the explainability concept. We develop for an orbital rendezvous scenario an innovative relative pose estimation technique adopting our proposed Convolutional Neural Network (CNN), which takes an image from the chaser's onboard camera and outputs accurately the target's relative position and rotation. We perturb seamlessly the input images using adversarial attacks that are generated by the Fast Gradient Sign Method (FGSM). The adversarial attack detector is then built based on a Long Short Term Memory (LSTM) network which takes the explainability measure namely SHapley Value from the CNN-based pose estimator and flags the detection of adversarial attacks when acting. Simulation results show that the proposed adversarial attack detector achieves a detection accuracy of 99.21%. Both the deep relative pose estimator and adversarial attack detector are then tested on real data captured from our laboratory-designed setup. The experimental results from our laboratory-designed setup demonstrate that the proposed adversarial attack detector achieves an average detection accuracy of 96.29%. [ABSTRACT FROM AUTHOR]

The short-term forecast of the critical frequency of the ionospheric F2 layer (foF2) is particularly challenging due to its nonlinear and non-stationary characteristics. To improve the short-term forecast accuracy of foF2, we propose a short-term hybrid deep learning model that integrates the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) and long short-term memory (LSTM) algorithms. The proposed model leverages the ICEEMDAN decomposition algorithm to expand foF2 time series data into multi-dimensional space and utilizes the LSTM algorithm to preserve long-term data information, capturing detailed changes in foF2 parameters and obtaining internal information about foF2. The 2014 and 2017 foF2 data from four observation stations located in Mohe, Beijing, Jeju, and Kokubunji in the mid-latitudes of East Asia were utilized for modeling and 1-hour forecast analysis. The forecast values of the proposed model align closely with the measured data, exhibiting minimal fluctuation in forecast error and effectively tracking the changing trend of foF2. The average root mean square error (RMSE) across the four stations in 2014 is 0.40 MHz, with a relative RMSE (RRMSE) of 7.11 % and a coefficient of determination (R2) of 0.98. The average RMSE in 2017 is 0.43 MHz, RRMSE is 7.68 %, and R2 is 0.92. The proposed model reveals significant improvements in the forecast accuracy of the comparison model compared with the CCIR, URSI, NN, and LSTM models. Especially compared with the CCIR and URSI models in the International Reference Ionosphere (IRI), demonstrating superior real-time tracking capability for foF2 and effectively addressing the challenge of poor short-term forecast accuracy observed in the IRI model. [ABSTRACT FROM AUTHOR]

One of the most important parameters for calculating the atmospheric precipitable water vapor (PWV) is the atmospheric weighted mean temperature (Tm). Meteorological parameters that are measured typically constrain the accuracy and regional applicability of commonly used empirical models. When precipitable water is inverted using the Global Navigation Satellite System (GNSS), it results in erroneous Tm being obtained. Thus, in this paper, a small-scale Tm model for Xi'an is proposed by least squares method based on data provided by the European Centre for Mesoscale Weather Forecasts (ECMWF) from 2019 to 2022, while improving the functional model's capacity to fit nonlinear residuals through the Random Forest method and enabling it to predict peaks more accurately. It is based on a random forest to confirm the weighted average temperature model's correctness (RF Tm). When comparing the RF Tm model's projected Tm accuracy to the GPT3, UNB3, and regional Bevis models, using the 2023 ERA5 data as the experimental data, the improvements are 23.6 %, 18.6 %, and 11.4 %, respectively. With the sounding station data as the reference value, the BIAS of RF Tm is only 1.51 K. In determining the atmospheric weighted average temperatures, the two models' accuracy was comparable to that of the Long Short-Term Memory neural network model (LSTM). However, the LSTM model requires short-term forward data for practical usage, the RF Tm model compensated for the shortfall. The RF Tm model is used to invert the atmospheric precipitable water vapor at the appropriate monitoring stations using data from GNSS monitoring stations in Xi'an. It was demonstrated that the RF Tm model's PWV inversion accuracy outperformed the PWV inverted by the GPT3 model by at least 30 %. Moreover, Xi'an rainfall data integration shows that the PWV inverted by the RF Tm model reacts to brief rainfall episodes efficiently. To sum up, this study's RF Tm model provides a dependable and effective approach to meteorological observation and climate research. [ABSTRACT FROM AUTHOR]

Enabling diabetic patients to predict their Blood Glucose Levels (BGL) is a crucial aspect of managing their metabolic condition, as it allows them to take appropriate measures to avoid hypo or hyperglycemia. Machine Learning (ML) and Deep Learning (DL) techniques have made this possible, and this paper evaluates and compares the performance of five distinct ML/DL models including: Convolutional Neural Network (CNN), Long Short Term Memory (LSTM), Support Vector Regression (SVR), Gated Reccurent Unit (GRU) and Deep Belief Network (DBN) for forecasting BGL, by applying two different forecasting methods, namely One Step Ahead (OSF) and Multi‐Step Ahead (MSF) comprising five different variants. The performance is evaluated based on four metrics: Mean Absolute Error (MAE), Mean Magnitude Relative Error (MMRE), Root Mean Square Error (RMSE) and Predictive Level (PRED). Additionally, the statistical significance of the regressors was evaluated using the Scott‐Knott (SK) test, while the Borda Count (BC) voting system was employed to rank them. The results indicate that the best performance was achieved with OSF using GRU. Furthermore, the effectiveness of an MSF strategy depends on the ML/DL technique used, and the best combinations were DBN with DirRec, DBN with Recursive, SVR with Recursive and SVR with DirRec. Additionally, DirRec was found to be the best strategy, as it consistently ranked first regardless of the ML/DL technique used. [ABSTRACT FROM AUTHOR]

In indoor positioning, Channel State Information (CSI) holds significant importance. CSI provides detailed channel characteristics, including amplitude and phase information, enabling the positioning system to estimate device locations more accurately. Compared to traditional RSSI methods, CSI offers higher positioning accuracy and better interference resistance. Additionally, it supports various advanced positioning algorithms, making it suitable for complex indoor environments and enabling real-time precise positioning. In this study, a localization method based on CSI phase and amplitude data is proposed with the aim of improving the accuracy and robustness of indoor localization. In the offline training phase, for each reference point, we construct a new dataset using pre-processed amplitude and phase data, which are transformed into image data for pre-training to extract features. Subsequently, a CNN-LSTM neural network with an integrated attention mechanism is trained to capture complex spatio-temporal relationships in the data. However, manual tuning of neural network hyperparameters has its limitations. Therefore, this paper introduces the Sparrow Search Algorithm (SSA)bio-inspired algorithm to optimize the combination of hyperparameters, aiming to obtain a more optimal model performance. During the online orientation phase, CSI data collected from the target device is processed using the same steps as in the offline phase to convert it into images. The trained model is then used to perform regression predictions on test points, enabling localization of the target position. To assess the effectiveness of our proposed approach, we ran tests in both open and complicated laboratory settings. Comparisons were made with several existing indoor localization solutions (such as PCNB, MIMO, LSTM, and FIFS), demonstrating that our approach exhibits significant advantages regarding localization accuracy and robustness. [ABSTRACT FROM AUTHOR]

Accurate near-real-time estimates of global horizontal irradiance (GHI) are crucial for understanding the Earth's energy budget and assessing the potential of solar energy as a renewable resource for mitigating climate change. However, dynamic atmospheric variations, such as aerosols and clouds, pose a challenge to accurate GHI estimation due to their unpredictable nature, leading to discrepancies between observed and simulated values, especially over short timeframes. To address this challenge, we combine the band information from the Himawari-8 satellite to enhance the ability to capture specific information such as clouds, aerosols, and atmospheric conditions. We further employ feature selection methods to input only the selected informative features into a hybrid model to mitigate possible shortcut learning. Our results show that the hybrid model (R²=0.90, RMSE = 88.38 W/m², RPD = 3.10, MAE = 57.75 W/m²) not only outperforms the deep neural network (DNN) (R²=0.82, RMSE = 116.78 W/m², RPD = 2.35, MAE = 80.47 W/m²) and the long short-term memory network (LSTM) (R²=0.87, RMSE = 97.94 W/m², RPD = 2.80, MAE = 65.05 W/m²) in terms of performance, but also exhibits good adaptability under different weather conditions. The results suggest that the use of a hybrid model helps balance the importance of various features, thereby reducing shortcut learning. This leads to more accurate and reliable estimates of GHI and provides new insights for obtaining high-frequency GHI over large spatial scales. [ABSTRACT FROM AUTHOR]

Forecasting of traffic in cellular network is a significant service for management of available resources strategically in an efficient way. Valuable resources such as link bandwidth and energy are increasing exponentially with increase in usage of cellular data. In this paper, we implement the design of Neural Network for identification of recurrent patterns in different metrics that can be then applied in forecasting of traffic in cellular networks. As this Neural Network design is based on memory and custom architecture, it is able to handle task of prediction in a precise and faster mode in real-time applications such as cellular network traffic forecasting. This work involves a Long Short Term Memory design of Recurrent Neural Network for traffic forecasting in cellular networks. It enhances the performance of the cellular network thereby providing a solution for the service providers as the available resources are utilized in an effective way. Same data set is involved for multiple prediction to analyze the performance of the design and found to be robust than the existing algorithms. [ABSTRACT FROM AUTHOR]

Paroxysmal kinesigenic dyskinesia (PKD) is a rare neurological disorder marked by transient involuntary movements triggered by sudden actions. Current diagnostic approaches, including genetic screening, face challenges in identifying secondary cases due to symptom overlap with other disorders. This study introduces a novel PKD recognition method utilizing a resting-state electroencephalogram (EEG) functional connectivity matrix and a deep learning architecture (AT-1CBL). Resting-state EEG data from 44 PKD patients and 44 healthy controls (HCs) were collected using a 128-channel EEG system. Functional connectivity matrices were computed and transformed into graph data to examine brain network property differences between PKD patients and controls through graph theory. Source localization was conducted to explore neural circuit differences in patients. The AT-1CBL model, integrating 1D-CNN and Bi-LSTM with attentional mechanisms, achieved a classification accuracy of 93.77% on phase lag index (PLI) features in the Theta band. Graph theoretic analysis revealed significant phase synchronization impairments in the Theta band of the functional brain network in PKD patients, particularly in the distribution of weak connections compared to HCs. Source localization analyses indicated greater differences in functional connectivity in sensorimotor regions and the frontal-limbic system in PKD patients, suggesting abnormalities in motor integration related to clinical symptoms. This study highlights the potential of deep learning models based on EEG functional connectivity for accurate and cost-effective PKD diagnosis, supporting the development of portable EEG devices for clinical monitoring and diagnosis. However, the limited dataset size may affect generalizability, and further exploration of multimodal data integration and advanced deep learning architectures is necessary to enhance the robustness of PKD diagnostic models. [ABSTRACT FROM AUTHOR]

Air pollution, particularly ambient particulate matter with aerodynamic diameter less than 2.5 μm (PM2.5), has emerged as a significant global concern due to its adverse impact on public health and the environment. Rapid urbanization, industrialization, and the increased number of automobiles in the cities have led to a significant enhancement in the PM2.5 concentrations to their hazardous level, which indicates the requirement for early warning systems to reduce exposure. Artificial Intelligence and Machine Learning (AI/ML) have come forth as highly sought-after tools widely utilized for air quality (AQ) forecasting. A deep learning based Recurrent Neural Network (RNN) models are highly being used due to their performance in predicting the AQ from the time series data. The present study evaluated three types of RNNs, namely SimpleRNN, Gradient Recurrent Units (GRU) and Long Short-Term Memory (LSTM) to forecast the PM2.5 in the four major Indian metropolitan cities. This research utilizes the daily in-situ PM2.5 data from national AQ monitoring agency in India, known as Central Pollution Control Board (CPCB) for the period 2018 to 2023. Various atmospheric gases and dispersion factors were employed to train model for the prediction of PM2.5 over the cities of Chennai, Delhi, Hyderabad and Kolkata. The ability of the each RNN model is evaluated and compared with observed data using various statistical parameters such as root mean squared error, mean absolute error, and mean absolute percentage error, coefficient of determination and correlation coefficient. Our findings indicate that all three neural networks can capture future PM2.5 trends consistently, albeit with some uncertainty. GRU was the most proficient in estimating PM2.5 levels in all the cities, followed by LSTM and SimpleRNN. The highest accuracy score was observed over Hyderabad followed by Kolkata, Chennai and Delhi. [ABSTRACT FROM AUTHOR]

The present investigation employs relevance vector machine (RVM) and long short-term memory (LSTM) models to predict the time-dependent bearing capacity of concrete piles. Each RVM model (SRVM) is configured by each linear, polynomial, gaussian, sigmoid, laplacian, and exponential kernel function. Each SRVM model has been optimized by each genetic (GA_SRVM) and particle swarm optimization (PSO_RVM) algorithm. Moreover, the double kernel-based RVM models (DRVM) have been employed and optimized by each GA (GA_DRVM) and PSO (PSO_DRVM) algorithm. Thus, an extensive comparison among 33 RVM (6SRVM + 6GA_RVM + 6PSO_RVM + 5DRVM + 5GA_DRVM + 5PSO_DRVM) has been carried out. Conversely, the Adam, root mean squared propagation and stochastic gradient descent with momentum algorithms have optimized the LSTM model. Each optimized RVM and LSTM model has been trained and tested by 100 and 26 datasets. In addition, the effect of structural and database multicollinearities has been analyzed on models' prediction capabilities. The performance index (PI), the variance accounted for (VAF), performance (R), mean absolute error (MAE), normalized mean bias error (NMBE), and root mean square error (RMSE) matrices have analyzed the prediction capabilities of each model. The comparison of 33 RVM and 3 LSTM models reveals that the genetic algorithm-optimized Gaussian kernel function-based SRVM model, i.e., UBC7, has been recognized as the optimal performance model with the RMSE = 146.3962 kPa, PI = 1.85, VAF = 94.60, NMBE = 30.1379 kPa, MAE = 105.7009 kPa, and R = 0.9727, close to the ideal values. Furthermore, the score (= 56), Wilcoxon (= 94.95% confidence), uncertainty (= 1st rank), generalizability (= close to ideal values), and Anderson Darling (= 9.435 ≈ 9.336) tests confirm the superiority of model UBC7. Still, structural and database multicollinearity has drastically impacted dual kernel-based RVM and stochastic gradient descent optimized LSTM models. [ABSTRACT FROM AUTHOR]

Parkinson's disease (PD) is a neurodegenerative disease affecting millions of people around the world. Conventional PD detection algorithms are generally based on first and second-generation artificial neural network (ANN) models which consume high energy and have complex architecture. Considering these limitations, a time-varying synaptic efficacy function based leaky-integrate and fire neuron model, called SEFRON is used for the detection of PD. SEFRON explores the advantages of Spiking Neural Network (SNN) which is suitable for neuromorphic devices. To evaluate the performance of SEFRON, 2 publicly available standard datasets, namely (1) UCI: Oxford Parkinson's Disease Detection Dataset and (2) UCI: Parkinson Dataset with replicated acoustic features are used. The performance is compared with other well-known neural network models: Multilayer Perceptron Neural Network (MLP-NN), Radial Basis Function Neural Network (RBF-NN), Recurrent Neural Network (RNN) and Long short-term memory (LSTM). The experimental results demonstrate that the SEFRON classifier achieves a maximum accuracy of 100% and an average accuracy of 99.49% on dataset 1. For dataset 2, it attains a peak accuracy of 94% and an average accuracy of 91.94%, outperforming the other classifiers in both cases. From the performance, it is proved that the presented model can help to develop a robust automated PD detection device that can assist the physicians to diagnose the disease at its early stage. [ABSTRACT FROM AUTHOR]

Long non-coding RNAs (lncRNAs) play crucial roles in numerous biological processes and are involved in complex human diseases through interactions with proteins. Accurate identification of lncRNA-protein interactions (LPI) can help elucidate the functional mechanisms of lncRNAs and provide scientific insights into the molecular mechanisms underlying related diseases. While many sequence-based methods have been developed to predict LPIs, efficiently extracting and effectively integrating potential feature information that reflects functional attributes from lncRNA and protein sequences remains a significant challenge. This paper proposes a Dinucleotide-Codon Fusion Feature encoding (DNCFF) and constructs an LPI prediction model based on deep learning, termed LPI-DNCFF. The Dual Nucleotide Visual Fusion Feature encoding (DNVFF) incorporates positional information of single nucleotides with subsequent nucleotide connections, while Codon Fusion Feature encoding (CFF) considers the specificity, molecular weight, and physicochemical properties of each amino acid. These encoding methods encapsulate rich and intuitive sequence information in limited encoding dimensions. The model comprehensively predicts LPIs by integrating global, local, and structural features, and inputs them into BiLSTM and attention layers to form a hybrid deep learning model. Experimental results demonstrate that LPI-DNCFF effectively predicts LPIs. The BiLSTM layer and attention mechanism can learn long-term dependencies and identify weighted key features, enhancing model performance. Compared to one-hot encoding, DNCFF more efficiently and thoroughly extracts potential sequence features. Compared to other existing methods, LPI-DNCFF achieved the best performance on the RPI1847 and ATH948 datasets, with MCC values of approximately 97.84% and 84.58%, respectively, outperforming the state-of-the-art method by about 1.44% and 3.48%. [ABSTRACT FROM AUTHOR]

As the primary power source for electric vehicles, the accurate estimation of the State of Health (SOH) of lithium-ion batteries is crucial for ensuring the reliable operation of the power system. Long Short-Term Memory (LSTM), a special type of recurrent neural network, achieves sequence information estimation through a gating mechanism. However, traditional LSTM-based SOH estimation methods do not account for the fact that the degradation sequence of battery SOH exhibits trend-like nonlinearity and significant dynamic variations between samples. Therefore, this paper proposes an LSTM-based lithium-ion SOH estimation method incorporating data characteristics and spatio-temporal attention. First, considering the trend-like nonlinearity of the degradation sequence, which is initially gradual and then rapid, input features are filtered and divided into trend and non-trend features. Then, to address the significant dynamic variations between samples, especially for capacity regeneration,a spatio-temporal attention mechanism is designed to extract spatio-temporal features from multidimensional non-trend features. Subsequently, an LSTM model is built with trend features, spatio-temporal features, and actual capacity as inputs to estimate capacity. Finally, the model is trained and tested on different datasets. Experimental results demonstrate that the proposed method outperforms traditional methods in terms of SOH estimation accuracy and robustness. [ABSTRACT FROM AUTHOR]

Sentiment Lexicon (SL) is utilized to extract feedback from a large amount of data, which encompass numerous words, in which a few words contain various semantic definitions in various fields, which is called domain-specific (DS) language, and every word under its domain is demonstrated very significantly as their definition varies from one another. Therefore, this research proposes a Self-configuring knowledge graph-based BiLSTM Classifier with Bat–Harris optimization to determine the accurate sentiment from the comment or text message as well as compute whether it is positive, negative, or neutral, by utilizing the feature extraction process which includes Term Frequency-Inverse Document Frequency (TF-IDF), as well as Hybrid word2vec features, and finally detect the sentiment polarity of text. The experimental results which are depending on the performance metrics show that the developed model is improved compared to the prior method, in which the rate of accuracy of the developed scheme on TP 90 is 93.40% and 94.29%, whereas the sensitivity attains the value of 93.31%, and 94.20%, and finally, the specificity acquires the value of 93.68%, and 94.66% based on datasets 1 and 2. [ABSTRACT FROM AUTHOR]

Background: Missing observations within the univariate time series are common in real-life and cause analytical problems in the flow of the analysis. Imputation of missing values is an inevitable step in every incomplete univariate time series. Most of the existing studies focus on comparing the distributions of imputed data. There is a gap of knowledge on how different imputation methods for univariate time series affect the forecasting performance of time series models. We evaluated the prediction performance of autoregressive integrated moving average (ARIMA) and long short-term memory (LSTM) network models on imputed time series data using ten different imputation techniques. Methods: Missing values were generated under missing completely at random (MCAR) mechanism at 10%, 15%, 25%, and 35% rates of missingness using complete data of 24-h ambulatory diastolic blood pressure readings. The performance of the mean, Kalman filtering, linear, spline, and Stineman interpolations, exponentially weighted moving average (EWMA), simple moving average (SMA), k-nearest neighborhood (KNN), and last-observation-carried-forward (LOCF) imputation techniques on the time series structure and the prediction performance of the LSTM and ARIMA models were compared on imputed and original data. Results: All imputation techniques either increased or decreased the data autocorrelation and with this affected the forecasting performance of the ARIMA and LSTM algorithms. The best imputation technique did not guarantee better predictions obtained on the imputed data. The mean imputation, LOCF, KNN, Stineman, and cubic spline interpolations methods performed better for a small rate of missingness. Interpolation with EWMA and Kalman filtering yielded consistent performances across all scenarios of missingness. Disregarding the imputation methods, the LSTM resulted with a slightly better predictive accuracy among the best performing ARIMA and LSTM models; otherwise, the results varied. In our small sample, ARIMA tended to perform better on data with higher autocorrelation. Conclusions: We recommend to the researchers that they consider Kalman smoothing techniques, interpolation techniques (linear, spline, and Stineman), moving average techniques (SMA and EWMA) for imputing univariate time series data as they perform well on both data distribution and forecasting with ARIMA and LSTM models. The LSTM slightly outperforms ARIMA models, however, for small samples, ARIMA is simpler and faster to execute. [ABSTRACT FROM AUTHOR]

Spreading fake news on social media is regarded as a cybercrime that greatly affects society, government and people. Identifying this fake news by manual analysis is a complex task. Hence, various researchers carried out an analysis on exploiting intelligent approaches to detect fake news. This work aims to propose a hybrid deep learning model for fake news detection. In these works, the news is verified in a unidirectional manner. Hence, there is a demand to change the present scenario and a new model is required for increasing the accuracy of the fake news detection. In this work, the deep learning (DL) model Long Short Term Memory (LSTM) with 3 Parallel-concatenated Convolutional Neural Networks (PCCNN) is utilised for fake news detection. For the feature extraction process, the methods like Term Frequency- Inverse Document Frequency (TF-IDF) and Latent Semantic Analysis (LSA) are utilised. Finally, for optimising the weights of the neural network, the metaheuristic optimisation Enhanced Sun Flower Optimisation (ESFO) algorithm is used. The robustness of LSTM-PCCNN is compared with other deep learning models to verify its robustness. The experimentation is carried out on the two benchmark datasets and obtained better accuracies of 0.994 on the fake news and 0.997 on the Real and fake news datasets respectively. [ABSTRACT FROM AUTHOR]

Recently, social demands for a good quality of life have increased among the elderly and disabled people. So, biomedical engineers and robotic researchers aimed to fuse these techniques in a novel rehabilitation system. Moreover, these models utilized the biomedical signals acquired from the human body's particular organ, cells, or tissues. The human motion intention prediction mechanism plays an essential role in various applications, such as assistive and rehabilitation robots, that execute specific tasks among elders and physically impaired individuals. However, more complications are increased in the human–machine-based interaction techniques, creating more scope for personalized assistance for the human motion intention prediction system. Therefore, in this paper, an Adaptive Hybrid Network (AHN) is implemented for effective human motion intention prediction. Initially, multimodal data like electroencephalogram (EEG)/Electromyography (EMG) signals and sensor measures data are collected from the available data resource. The gathered EEG/EMG signals are then converted into spectrogram images and sent to AH-CNN-LSTM, which is the integration of an Adaptive Hybrid Convolution Neural Network (AH-CNN) with a Long Short-Term Memory (LSTM) network. Similarly, the data details of sensor measures are directly subjected to AH-CNN-Res-LSTM, which is the combination of Adaptive Hybrid CNN with Residual Network and LSTM (Res-LSTM) to get the predictive result. Further, to enhance the prediction, the parameters in both the AH-CNN-LSTM and AH-CNN-Res-LSTM techniques are optimized using the Improved Yellow Saddle Goatfish Algorithm (IYSGA). The efficiency of the implemented model is computed by conducting the comparison experiment of the proposed technique with other standard models. The performance outcome of the developed method outperformed the other traditional methods. [ABSTRACT FROM AUTHOR]

Arabic dialect identification (ADI) is a specific task of natural language processing (NLP) that intends to forecast the Arabic language dialect of the input text automatically. ADI is the preliminary step toward establishing many NLP applications, including cross-language text generation, multilingual text-to-speech synthesis, and machine translation. The automatic classification of the Arabic dialect is the first step in various dialect-sensitive Arabic NLP tasks. ADI includes predicting the dialects related to the textual input and classifying them on their respective labels. As a result, increased interest has been gained in the last few decades to address the problems of ADI through deep learning (DL) and machine learning (ML) algorithms. The study develops an Arabic multi-class dialect recognition using fast random opposition-based fractals learning aquila optimizer with DL (FROBLAO-DL) technique. The FROBLAO-DL technique utilizes the optimal DL model to identify distinct types of Arabic dialects. In the FROBLAO-DL technique, data preprocessing is involved in cleaning the input Arabic dialect dataset. In addition, the ROBERTa word embedding process is used to generate word embedding. The FROBLAO-DL technique uses attention bidirectional long short-term memory (ABiLSTM) network to identify distinct Arabic dialects. Moreover, the ABiLSTM model’s hyperparameter tuning is implemented using the FROBLOA method. The performance evaluation of the FROBLAO-DL method is tested under the Arabic dialect dataset. The empirical analysis implies the supremacy of the FROBLAO-DL technique over recent approaches under various measures. [ABSTRACT FROM AUTHOR]

The conversion of handwritten text into machine-readable format is termed as Handwritten Character Recognition (HCR). The differences in size, design, and alignment angle of the Telugu and Kannada alphabets have difficulty in recognizing handwritten documents in these languages across various real-world applications. Newly developed machine learning and deep learning models provide a significant improvements in the handwritten text recognition. These innovative methods offer promising enhancements in the accuracy and efficiency of character recognition within handwritten documents. However, effective recognition of digits is not an easy task due to people’s varying writing styles in the input sample. To overcome such limitations, we explore a novel approach specifically designed to boost the performance of HCR in South Indian languages such as Kannada and Telugu. Initially, handwritten images are gathered using traditional data sources. These collected images are then given into the recognition phase. Here, an Adaptive Dilated convolution-based Deep Network (ADC-DeepNet) is developed for character identification purposes. In ADC-DeepNet, the ShuffleNetV2 blends with the Bidirectional Long Short-Term Memory (Bi-LSTM) to produce accurate results. This fusion provides effective character recognition. Here, the Iterative Concept of Lyrebird Optimization (ICLO) is newly proposed to optimize the variables from ADC-DeepNet to improve the character recognition efficacy. The efficiency of the HCR system is evaluated among several recent techniques with some performance measures. Finally, the outcome showed that the accuracy of the proposed approach is 95.6, and other models like CNN, ResNet, Convolutional Autoencoder, and DeepNet gave the accuracy of 88.8, 91.5, 90.6, and 93.3, respectively. Thus, the findings of the experiment show that the developed ADC-DeepNet model can effectively identify the handwritten characters in south Indian languages. [ABSTRACT FROM AUTHOR]

Pterocarpus santalinus L.f. (P. santalinus), protected under the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES), is a high-priced, slow-growing, and scarce wood primarily used in crafting high-end furniture. The international timber trade currently faces issues of counterfeit P. santalinus, with commonly used substitutes including Dalbergia louvelii R.Viguier, Pterocarpus tinctorius Welw., Gluta renghas L. and Baphia nitida Lodd. This study aims to develop a P. santalinus authenticity identification model based on near-infrared spectroscopy (NIRS) technology. The NIR spectral pretreatment involved the use of four methods, either individually or in combination: multiplicative scatter correction (MSC), moving average smoothing (MAS), Savitzky-Golay (S-G), autoscaling (AUTO) and standard normal variate (SNV). An authenticity identification model for P. santalinus based on long short-term memory (LSTM) was established and compared with commonly used support vector machines (SVM) and random forest (RF) models. The results indicate that the accuracy of the MSC-LSTM model is 97.1 %, with precision, recall, and F1 score all exceeding 0.85. In identifying P. santalinus in the test set, the MSC-LSTM model has an error rate of only 4.8 %. LSTM performs outstandingly across multiple indicators, demonstrating its ability to identify P. santalinus authenticity. The developed MSC-LSTM P. santalinus authenticity identification model shows enhanced accuracy compared to SVM and RF, significantly reducing misidentification of P. santalinus. [ABSTRACT FROM AUTHOR]

This study proposes an AI-based framework to predict the quality of results from the deep drawing process without requiring considerable amounts of actual data. A few-shot learning method with a Siamese network was employed for training with limited experimental data and virtual simulation data. As the base of the Siamese network, three network architectures, fully connected network (FCN), convolutional neural network (CNN), and long short-term memory (LSTM), were considered. By increasing the ratio of the experimental data, the best accuracy for each network was determined. Finally, the K-nearest neighbor algorithm was trained to predict the drawing quality using the embedding vector from the Siamese network. In all networks, it was confirmed that the accuracy of the classification algorithm improved as more experimental data were added. At this time, the classification accuracy of the CNN was 100%, which was significantly superior to the 77.8% of the FCN and 88.9% of the LSTM. Finally, the AI-based framework proposed in this paper can be effectively applied to various press forming processes as a method for predicting the quality of deep drawing molded products. [ABSTRACT FROM AUTHOR]

Image processing and pattern recognition methods have recently been extensively implemented in histopathological images (HIs). These computer-aided techniques are aimed at detecting the attentive biological markers for assisting the final cancer grading. Mitotic count (MC) is a significant cancer detection and grading parameter. Conventionally, a pathologist examines the biopsy image physically by employing higher-power microscopy. The MC cells have been marked physically at every analysis, and total MC must be utilized as a major aspect for the cancer ranking and considered as the initiative of cancers. Numerous pattern recognition algorithms for cell-sized objects in HIs depend upon segmentation to assess features. The correct description of the segmentation has been difficult, and feature outcomes can be highly complex to the segmentation. The MC cells are an essential element in many cancer grading methods. Extraction of the MC cell from the HI is a highly challenging assignment. This manuscript proposes the Coati Optimization Algorithm with Deep Learning-Driven Mitotic Nuclei Segmentation and Classification (COADL-MNSC) technique. The major aim of the COADL-MNSC technique is to utilize the DL model to segment and classify the mitotic nuclei (MN). In the preliminary stage, the COADL-MNSC approach implements median filtering (MF) for pre-processing. Besides, the COADL-MNSC approach utilizes the Hybrid Attention Fusion U-Net (HAU-UNet) model to segment the MN. Moreover, the capsule network (CapsNet) model is employed for the feature extraction method, and its hyperparameters are adjusted by utilizing the COA model. At last, the classification procedure is performed using the bidirectional long short-term memory (BiLSTM) model. Extensive simulations are performed under the MN image dataset to exhibit the excellent performance of the COADL-MNSC methodology. The experimental validation of the COADL-MNSC methodology portrayed a superior accuracy value of 98.89% over existing techniques under diverse measures. [ABSTRACT FROM AUTHOR]

Accurate travel time information is essential for logistics vehicles to reserve the most suitable parking lot in logistics centers. The purpose of this study is to explore how the uncertainty of traffic time prediction affects parking lot reservation near logistics centers. A hybrid model integrating convolutional long short-term memory network and attention mechanism is proposed to provide the reliable information for travel time prediction intervals. Furthermore, a reliability-based parking lot reservation model is developed by explicitly considering logistics vehicles' time probability. Several benchmark models are compared with the proposed traffic speed prediction model. The performance of the parking lot reservation model is illustrated by travel behavior questionnaire data and global positioning system data of collected from Beijing, China. The results illustrate that the proposed prediction model exhibits a better accuracy than benchmark models. Moreover, it is found that travel time prediction interval can improve the reliability and stability of travel time, and provide a reliable time information for the parking lot reservation. [ABSTRACT FROM AUTHOR]

Informal education via social media plays a crucial role in modern learning, offering self-directed and community-driven opportunities to gain knowledge, skills, and attitudes beyond traditional educational settings. These platforms provide access to a broad range of learning materials, such as tutorials, blogs, forums, and interactive content, making education more accessible and tailored to individual interests and needs. However, challenges like information overload and the spread of misinformation highlight the importance of digital literacy in ensuring users can critically evaluate the credibility of information. Consequently, the significance of sentiment analysis has grown in contemporary times due to the widespread utilization of social media platforms as a means for individuals to articulate their viewpoints. Twitter (now X) is well recognized as a prominent social media platform that is predominantly utilized for microblogging. Individuals commonly engage in expressing their viewpoints regarding contemporary events, hence presenting a significant difficulty for scholars to categorize the sentiment associated with such expressions effectively. This research study introduces a highly effective technique for detecting misinformation related to the COVID-19 pandemic. The spread of fake news during the COVID-19 pandemic has created significant challenges for public health and safety because misinformation about the virus, its transmission, and treatments has led to confusion and distrust among the public. This research study introduce highly effective techniques for detecting misinformation related to the COVID-19 pandemic. The methodology of this work includes gathering a dataset comprising fabricated news articles sourced from a corpus and subjected to the natural language processing (NLP) cycle. After applying some filters, a total of five machine learning classifiers and three deep learning classifiers were employed to forecast the sentiment of news articles, distinguishing between those that are authentic and those that are fabricated. This research employs machine learning classifiers, namely Support Vector Machine, Logistic Regression, K-Nearest Neighbors, Decision Trees, and Random Forest, to analyze and compare the obtained results. This research employs Convolutional Neural Networks, Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU) as deep learning classifiers, and afterwards compares the obtained results. The results indicate that the BiGRU deep learning classifier demonstrates high accuracy and efficiency, with the following indicators: accuracy of 0.91, precision of 0.90, recall of 0.93, and F1-score of 0.92. For the same algorithm, the true negatives, and true positives came out to be 555 and 580, respectively, whereas, the false negatives and false positives came out to be 81, and 68, respectively. In conclusion, this research highlights the effectiveness of the BiGRU deep learning classifier in detecting misinformation related to COVID-19, emphasizing its significance for fostering media literacy and resilience against fake news in contemporary society. The implications of this research are significant for higher education and lifelong learners as it highlights the potential for using advanced machine learning to help educators and institutions in the process of combating the spread of misinformation and promoting critical thinking skills among students. By applying these methods to analyze and classify news articles, educators can develop more effective tools and curricula for teaching media literacy and information validation, equipping students with the skills needed to discern between authentic and fabricated information in the context of the COVID-19 pandemic and beyond. The implications of this research extrapolate to the creation of a society that is resistant to the spread of fake news through social media platforms. [ABSTRACT FROM AUTHOR]

Background: Interactions between microRNAs and RNA-binding proteins are crucial for microRNA-mediated gene regulation and sorting. Despite their significance, the molecular mechanisms governing these interactions remain underexplored, apart from sequence motifs identified on microRNAs. To date, only a limited number of microRNA-binding proteins have been confirmed, typically through labor-intensive experimental procedures. Advanced bioinformatics tools are urgently needed to facilitate this research. Methods: We present DeepMiRBP, a novel hybrid deep learning model specifically designed to predict microRNA-binding proteins by modeling molecular interactions. This innovation approach is the first to target the direct interactions between small RNAs and proteins. DeepMiRBP consists of two main components. The first component employs bidirectional long short-term memory (Bi-LSTM) neural networks to capture sequential dependencies and context within RNA sequences, attention mechanisms to enhance the model's focus on the most relevant features and transfer learning to apply knowledge gained from a large dataset of RNA-protein binding sites to the specific task of predicting microRNA-protein interactions. Cosine similarity is applied to assess RNA similarities. The second component utilizes Convolutional Neural Networks (CNNs) to process the spatial data inherent in protein structures based on Position-Specific Scoring Matrices (PSSM) and contact maps to generate detailed and accurate representations of potential microRNA-binding sites and assess protein similarities. Results: DeepMiRBP achieved a prediction accuracy of 87.4% during training and 85.4% using testing, with an F score of 0.860. Additionally, we validated our method using three case studies, focusing on microRNAs such as miR-451, -19b, -23a, -21, -223, and -let-7d. DeepMiRBP successfully predicted known miRNA interactions with recently discovered RNA-binding proteins, including AGO, YBX1, and FXR2, identified in various exosomes. Conclusions: Our proposed DeepMiRBP strategy represents the first of its kind designed for microRNA-protein interaction prediction. Its promising performance underscores the model's potential to uncover novel interactions critical for small RNA sorting and packaging, as well as to infer new RNA transporter proteins. The methodologies and insights from DeepMiRBP offer a scalable template for future small RNA research, from mechanistic discovery to modeling disease-related cell-to-cell communication, emphasizing its adaptability and potential for developing novel small RNA-centric therapeutic interventions and personalized medicine. [ABSTRACT FROM AUTHOR]

Efficiently determining a price and optimal exercise boundary for an American option is a critical subject in the financial sector. This study introduces a novel application of long short-term memory neural networks to solve a relevant Volterra equation, enhancing the accuracy and efficiency of American option pricing. The proposed approach outperforms traditional numerical techniques, including finite difference methods, binomial trees, and Monte Carlo methods, delivering an impressive speed improvement by a factor of thousands while maintaining industry-accepted accuracy levels. It exhibits computational speeds up to about a hundred times faster than the state-of-the-art method used by Andersen et al. [High-performance American option pricing, J. Comput. Finance 20(1) (2016), pp. 39–87] when evaluating numerous options. The proposed network, trained on a reasonable range of parameters related to the Black–Scholes model, swiftly determines option prices and exercise boundaries, serving as a practical closed-form solution. [ABSTRACT FROM AUTHOR]

This study proposes a novel multi-agent system designed to detect Distributed Denial of Service (DDoS) attacks, addressing the increasing need for robust cybersecurity measures. The hypothesis posits that a structured multi-agent approach can enhance detection accuracy and response efficiency in DDoS attack scenarios. The methodology involves a five-stage detection model: (1) Preprocessing using a modified double sigmoid normalization technique to eliminate duplicate data; (2) Feature Extraction where raw data and improved correlation-based features, mutual information, and statistical features are identified; (3) Dimensionality Reduction conducted by a reducer agent to streamline the feature set; (4) Classification utilizing Deep Belief Networks (DBN), Bi-LSTM, and Deep Maxout models, with their weights optimally tuned using the hybrid optimization algorithm, WUJSO; and (5) Decision Making by the decision agent to ascertain the presence of attacks, followed by mitigation through modified entropy-based techniques. The results demonstrate that the proposed method achieves a detection accuracy of 0.953 at a learning rate of 90%, significantly outperforming other methods, including Bi-GRU (0.857), DEEP-MAXOUT (0.910), Bi-LSTM (0.865), RNN (0.814), NN (0.894), and DBN (0.761). This research underscores the effectiveness of the multi-agent approach in enhancing DDoS attack detection and mitigation. [ABSTRACT FROM AUTHOR]

In this study, we propose an innovative latent space-based machine learning model to predict coupled flame-flow fields across various combustion states. This model consists of two main components: a local convolutional autoencoder (LCAE) with an encoder to establish a latent space and a decoder to restore predicted features to their original physical dimensions, and a long short-term memory (LSTM) model to forecast the spatiotemporal behavior of features in this latent space. The model is trained and validated using flame-flow images simultaneously measured in a gas turbine model combustor, utilizing synchronized stereoscopic particle image velocimetry (S-PIV) and planar laser induced fluorescence of the hydroxyl radical (OH-PLIF). Our model effectively extracts large-scale structures from the coupled flame-flow measurements, even in the presence of aleatoric noise, and demonstrates satisfactory prediction and generalization capabilities, as verified through out-of-sample testing. By establishing a common latent space for the coupled flame-flow fields, we reveal inherent similarities between flame and flow fields in a low-dimensional feature space, suggesting the potential for holistic understanding and modeling of their behavior in such a space. Additionally, our model predicts rapidly and reliably, requiring only 403 ms for out-of-sample predictions, with no error accumulation across 499 predictions. This paves the way for early warning of impending undesirable combustion states in future applications. • A novel machine learning model for predicting coupled flame-flow fields is proposed. • The model framework enables rapid, stable predictions with a simple workflow. • The model shows strong generalization and robustness with low computational cost. [ABSTRACT FROM AUTHOR]

Reservoir pressure depletion is a critical aspect leading to declined oil production. Water flooding becomes mandatory to increase the recovery efficiency from certain reservoirs. Establishing well placements, injection rates, and injection schedules enables improved fluid displacement in such reservoirs. Standard reservoir simulators are widely used to provide near-optimal scenarios for maximizing oil recovery. Such reservoir simulations involved numerous iterations and required significant computational effort. Also, the conventional machine learning methods provide a relatively high uncertainty to predict the considered problem. In this context, a deep-learning surrogate model comprising long short-term memory networks and densely connected networks is proposed to predict cumulative oil production using various waterflood injection well patterns, injection rates, and injection period information. The relevance test clarifies that the oil recovery mainly controls by water injection duration and injection pattern. A total of 1485 data records are considered to validate the model's performance, demonstrating less than 3.5% absolute average relative error and a prediction versus actual value regression coefficient of 0.99989. In addition, the prediction accuracy of the proposed deep-learning model is also shown to outperform two well-known traditional machine-learning models (i.e. artificial neural networks and support vector regressions). [ABSTRACT FROM AUTHOR]

Crude oil is a raw and natural, but nonrenewable, resource. It is one of the world's most important commodities, and its price can have ripple effects throughout the broader economy. Accurately predicting crude oil prices is vital for investment decisions but it remains challenging. Due to the deficiencies neglecting residual factors when forecasting using conventional combination models, such as the autoregressive moving average and the long short-term memory for prediction, the variational mode decomposition (VMD)-empirical mode decomposition (EMD)-Transformer model is proposed to predict crude oil prices in this study. This model integrates a second decomposition and Transformer model-based machine learning method. More specifically, we employ the VMD technique to decompose the original sequence into variational mode filtering (VMF) and a residual sequence, followed by using EMD to decompose the residual sequence. Ultimately, we apply the Transformer model to predict the decomposed modal components and superimpose the results to produce the final forecasted prices. Further empirical test results demonstrate that the proposed quadratic decomposition composite model can comprehensively identify the characteristics of WTI and Brent crude oil futures daily price series. The test results illustrate that the proposed VMD–EMD–Transformer model outperforms the other three models—long short-term memory (LSTM), Transformer, and VMD–Transformer in forecasting crude oil prices. Details are presented in the empirical study part. [ABSTRACT FROM AUTHOR]

Electric shock protection is critical for ensuring power safety in low-voltage grids, and robust fault diagnosis methods provide an essential foundation for the accurate operation of such protection devices. However, current low-voltage electric shock protection devices often suffer from limitations in operational precision and in their ability to effectively recognize electric shock types. To address these challenges, this paper proposes a fault diagnosis method for low-voltage electric shocks based on an attention-enhanced parallel CNN-BiLSTM model. The method first utilizes CNN to extract local spatial features of the electric shock signal and BiLSTM to capture temporal features. An attention mechanism is then introduced to fuse the local spatial and temporal features with weighted emphasis. Finally, a fully connected layer maps the fused features to the output layer, generating diagnostic results. Visualization through T-SNE analysis validates the improvement in model performance due to the attention mechanism. Comparative experiments show that the proposed model outperforms single models and other combined models in terms of accuracy, precision, recall, F1 score, and convergence speed. The results demonstrate that the proposed model achieves a fault diagnosis accuracy of 99.55%. [ABSTRACT FROM AUTHOR]

This study introduces a novel methodology that integrates the Black–Litterman model with Long Short-Term Memory Neural Networks (BL–LSTM). We use predictions from the LSTM as views in the Black–Litterman model. The resulting portfolio performs better than the traditional mean-variance (MV) and exchange-traded funds (ETFs) used as benchmarks. The proposal empowers investors to make more insightful decisions, drawing from a synthesis of historical data and advanced predictive techniques. This methodology is applied to a water market. Investing in the water market allows investors to actively support sustainable water solutions while potentially benefiting from the sector's growth, contributing to achieving SDG 6. In addition, our modeling allows for companies' environmental, social, and governance (ESG) scores to be considered in the portfolio construction process. In this case, investors' decisions take into account companies' socially responsible behavior in a broad sense, including aspects related to decent work, respect for indigenous communities and diversity, and the absence of corruption, among others. Therefore, this proposal provides investors with a tool for promoting sustainable investment practices. [ABSTRACT FROM AUTHOR]

Financial time series prediction is a fundamental problem in investment and risk management. Deep learning models, such as multilayer perceptrons, Convolutional Neural Networks (CNNs), and Long Short-Term Memory (LSTM), have been widely used in modeling time series data by incorporating historical information. Among them, LSTM has shown excellent performance in capturing long-term temporal dependencies in time-series data, owing to its enhanced internal memory mechanism. In spite of the success of these models, it is observed that in the presence of sharp changing points, these models fail to perform. To address this problem, we propose, in this article, an innovative financial time series prediction method inspired by the Deep Operator Network (DeepONet) architecture, which uses a combination of transformer architecture and a one-dimensional CNN network for processing feature-based information, followed by an LSTM based network for processing temporal information. It is therefore named the CNN–LSTM–Transformer (CLT) model. It not only incorporates external information to identify latent patterns within the financial data but also excels in capturing their temporal dynamics. The CLT model adapts to evolving market conditions by leveraging diverse deep-learning techniques. This dynamic adaptation of the CLT model plays a pivotal role in navigating abrupt changes in the financial markets. Furthermore, the CLT model improves the long-term prediction accuracy and stability compared with state-of-the-art existing deep learning models and also mitigates adverse effects of market volatility. The experimental results show the feasibility and superiority of the proposed CLT model in terms of prediction accuracy and robustness as compared to existing prediction models. Moreover, we posit that the innovation encapsulated in the proposed DeepONet-inspired CLT model also holds promise for applications beyond the confines of finance, such as remote sensing, data mining, natural language processing, and so on. [ABSTRACT FROM AUTHOR]

Recent advances in emotion recognition through Artificial Intelligence (AI) have demonstrated potential applications in various fields (e.g., healthcare, advertising, and driving technology), with electroencephalogram (EEG)-based approaches demonstrating superior accuracy compared to facial or vocal methods due to their resistance to intentional manipulation. This study presents a novel approach to enhance EEG-based emotion estimation accuracy by emphasizing temporal features and efficient parameter space exploration. We propose a model combining Long Short-Term Memory (LSTM) with an attention mechanism to highlight temporal features in EEG data while optimizing LSTM parameters through Particle Swarm Optimization (PSO). The attention mechanism assigned weights to LSTM hidden states, and PSO dynamically optimizes the vital parameters, including units, batch size, and dropout rate. Using the DEAP and SEED datasets, which serve as benchmark datasets for emotion estimation research using EEG, we evaluate the model's performance. For the DEAP dataset, we conduct a four-class classification of combinations of high and low valence and arousal states. We perform a three-class classification of negative, neutral, and positive emotions for the SEED dataset. The proposed model achieves an accuracy of 0.9409 on the DEAP dataset, surpassing the previous state-of-the-art accuracy of 0.9100 reported by Lin et al. The model attains an accuracy of 0.9732 on the SEED dataset, recording one of the highest accuracies among the related research. These results demonstrate that integrating the attention mechanism with PSO significantly improves the accuracy of EEG-based emotion estimation, contributing to the advancement of emotion recognition technology. [ABSTRACT FROM AUTHOR]

Satellite SAR (synthetic aperture radar) imagery offers global coverage and all-weather recording capabilities, making it valuable for applications like remote sensing and maritime surveillance. However, its use in machine learning-based automatic target classification faces challenges, including the limited availability of SAR target training samples and the inherent constraints of SAR images, which provide less detailed features compared to natural images. These issues hinder the effective training of convolutional neural networks (CNNs) and complicate the transfer learning process due to the distinct imaging mechanisms of SAR and natural images. To address these challenges, we propose a shallow CNN architecture specifically designed to optimize performance on SAR datasets. Evaluations were performed on three datasets: FUSAR-Ship, OpenSARShip, and MSTAR. While the FUSAR-Ship and OpenSARShip datasets present difficulties due to their limited and imbalanced class distributions, MSTAR serves as a benchmark with balanced classes. To compare and optimize the proposed shallow architecture, we examine various properties of CNN components, such as the filter numbers and sizes in the convolution layers, to reduce redundancy, improve discrimination capability, and decrease network size and learning time. In the second phase of this paper, we combine the CNN with Long short-term memory (LSTM) networks to enhance SAR image classification. Comparative experiments with six state-of-the-art CNN architectures (VGG16, ResNet50, Xception, DenseNet121, EfficientNetB0, and MobileNetV2) demonstrate the superiority of the proposed approach, achieving competitive accuracy while significantly reducing training times and network complexity. This study underscores the potential of customized architectures to address SAR-specific challenges and enhance the efficiency of target classification. [ABSTRACT FROM AUTHOR]

Intrusion Detection Systems (IDSs) are a crucial component of modern corporate firewalls. The ability of IDS to identify malicious traffic is a powerful tool to prevent potential attacks and keep a corporate network secure. In this context, Machine Learning (ML)-based methods have proven to be very effective for attack identification. However, traditional approaches are not always applicable in a real-time environment as they do not integrate concrete traffic management after a malicious packet pattern has been identified. In this paper, a novel combined approach to both identify and discard potential malicious traffic in a real-time fashion is proposed. In detail, a Long Short-Term Memory (LSTM) supervised artificial neural network model is provided in which consecutive packet groups are considered as they flow through the corporate network. Moreover, the whole IDS architecture is modeled by a Petri Net (PN) that either blocks or allows packet flow throughout the network based on the LSTM model output. The novel hybrid approach combining LSTM with Petri Nets achieves a 99.71% detection accuracy—a notable improvement over traditional LSTM-only methods, which averaged around 97%. The LSTM–Petri Net approach is an innovative solution combining machine learning with formal network modeling for enhanced threat detection, offering improved accuracy and real-time adaptability to meet the rapid security needs of virtual environments and CPS. Moreover, the approach emphasizes the innovative role of the Intrusion Detection System (IDS) and Intrusion Prevention System (IPS) as a form of "virtual sensing technology" applied to advanced network security. An extensive case study with promising results is provided by training the model with the popular IDS 2018 dataset. [ABSTRACT FROM AUTHOR]

Sensor networks generate vast amounts of data in real-time, which challenges existing predictive maintenance frameworks due to high latency, energy consumption, and bandwidth requirements. This research addresses these limitations by proposing an edge-cloud hybrid framework, leveraging edge devices for immediate anomaly detection and cloud servers for in-depth failure prediction. A K-Nearest Neighbors (KNNs) model is deployed on edge devices to detect anomalies in real-time, reducing the need for continuous data transfer to the cloud. Meanwhile, a Long Short-Term Memory (LSTM) model in the cloud analyzes time-series data for predictive failure analysis, enhancing maintenance scheduling and operational efficiency. The framework's dynamic workload management algorithm optimizes task distribution between edge and cloud resources, balancing latency, bandwidth usage, and energy consumption. Experimental results show that the hybrid approach achieves a 35% reduction in latency, a 28% decrease in energy consumption, and a 60% reduction in bandwidth usage compared to cloud-only solutions. This framework offers a scalable, efficient solution for real-time predictive maintenance, making it highly applicable to resource-constrained, data-intensive environments. [ABSTRACT FROM AUTHOR]

This paper examines Connected Smart Green Buildings (CSGBs) in Burnaby, BC, Canada, with a focus on townhouses with one to four bedrooms. The proposed model integrates sustainable materials and smart components such as recycled insulation, Photovoltaic (PV) solar panels, smart meters, and high-efficiency systems. These elements improve energy efficiency and promote sustainability. Operating in island mode, CSGBs can function independently of the grid, providing resilience during power outages and reducing reliance on external energy sources. Real data on electricity, gas, and water consumption are used to optimize load management under isolated conditions. Electric Vehicles (EVs) are also considered in the system. They serve as energy storage devices and, through Vehicle-to-Grid (V2G) technology, can supply power when needed. A hybrid Machine Learning (ML) model combining Long Short-Term Memory (LSTM) and a Convolutional Neural Network (CNN) is proposed to improve the performance. The metrics considered include accuracy, efficiency, emissions, and cost. The performance was compared with several well-known models including Linear Regression (LR), CNN, LSTM, Random Forest (RF), Gradient Boosting (GB), and hybrid LSTM–CNN, and the results show that the proposed model provides the best results. For a four-bedroom Connected Smart Green Townhouse (CSGT), the Mean Absolute Percentage Error (MAPE) is 4.43%, the Root Mean Square Error (RMSE) is 3.49 kWh, the Mean Absolute Error (MAE) is 3.06 kWh, and R 2 is 0.81. These results indicate that the proposed model provides robust load optimization, particularly in island mode, and highlight the potential of CSGBs for sustainable urban living. [ABSTRACT FROM AUTHOR]

In this paper, we propose the XPaC (XGBoost Prediction and Cumulative Prospect Theory (CPT)) model to minimize the operational losses of the power grid, taking into account both the prediction of electric vehicle (EV) charging demand and the associated uncertainties, such as when customers will charge, how much electric energy they will need, and for how long. Given that power utilities supply electricity with limited resources, it is crucial to efficiently control EV charging peaks or predict charging demand during specific periods to maintain stable grid operations. While the total amount of EV charging is a key factor, when and where the charging occurs can be even more critical for the effective management of the grid. Although numerous studies have focused on individually predicting EV charging patterns or demand and evaluating the effectiveness of EV charging control, comprehensive assessments of the actual operational benefits and losses resulting from charging control based on predicted charging behavior remain limited. In this study, we firstly compare the performance of LSTM (Long Short-Term Memory), GRU (Gated Recurrent Unit), and decision tree-based XGBoost regression models in predicting hourly charging probabilities and the need for grid demand control. Using the predicted results, we applied the CPT algorithm to analyze the optimal operational scenarios and assess the expected profit and loss for the power grid. Since the charging control optimizer with XPaC incorporates real-world operational data and uses actual records for analysis, it is expected to provide a robust solution for managing the demand arising from the rapid growth of electric vehicles, while operating within the constraints of limited energy resources. [ABSTRACT FROM AUTHOR]

Accurate energy consumption prediction in commercial buildings is a challenging research task. Energy prediction plays a crucial role in energy efficiency, management, planning, sustainability, risk management, diagnosis, and demand response. Although many studies have been conducted on building energy predictions, the impact of occupancy on energy prediction models for office-type commercial buildings remains insufficiently explored, despite its potential to improve energy efficiency by 20%. This study investigates energy prediction using a Long Short-Term Memory (LSTM) model that incorporates time-series power consumption data and considers occupancy. A real-world dataset containing the per-minute electricity consumption of various appliances in an office building in Houston, TX, USA, is utilized. The proposed machine learning models forecast future energy consumption based on hourly, 3-hourly, daily, and quarterly predictions for individual appliances and total energy usage. The model's performance is evaluated using the following three metrics: Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). The results demonstrate the superiority of the proposed system. [ABSTRACT FROM AUTHOR]

Vehicle-to-grid (V2G) technology is emerging as an innovative paradigm for improving the electricity grid in terms of stabilization and demand response, through the integration of electric vehicles (EVs). A cornerstone in this field is the estimation of the aggregated available capacity (AAC) of EVs based on available data such as origin–destination mobility data, traffic and time of day. This paper considers a real case study, consisting of two aggregation points, identified in the city of Padua (Italy). As a result, this study presents a new method to identify potential applications of V2G by analyzing floating car data (FCD), which allows planners to infer the available AAC obtained from private vehicles. Specifically, the proposed method takes advantage of the opportunity provided by FCD to find private car users who may be interested in participating in V2G schemes, as telematics and location-based applications allow vehicles to be continuously tracked in time and space. Linear and nonlinear dynamic models with different input variables were developed to analyze their relevance for the estimation in one-step- and multiple-step-ahead prediction. The best results were obtained by using traffic data as exogenous input and nonlinear dynamic models implemented by multilayer perceptrons and long short-term memory (LSTM) networks. Both structures achieved an R 2 of 0.95 and 0.87 for the three-step-ahead AAC prediction in the two hubs considered, compared to the values of 0.88 and 0.72 obtained with the linear autoregressive model. In addition, the transferability of the obtained models from one aggregation point to another was analyzed to address the problem of data scarcity in these applications. In this case, the LSTM showed the best performance when the fine-tuning strategy was considered, achieving an R 2 of 0.80 and 0.89 for the two hubs considered. [ABSTRACT FROM AUTHOR]