Search

Road traffic management requires the ability to foresee geographical congestion conditions in an urban road traffic network. The proposed investigation is aimed to envisage the presence of blockage in a specific region of geographical location using gated recurrent unit (GRU) combined with the convolutional neural network (CNN) and long short-term memory (LSTM). The data related to road traffic is extracted from the time—frequency domain of a specific urban road. The collected data is then compressed into a RGB coded image which is the input to the GRU which determines the parameters of traffic blockage in a specific region of the road. The road traffic network features are extracted from those parameters and identified using a CNN's feature extraction technology. We used LSTM model to assess the time series temporal images of the road traffic blockage and to utilize the time division to pool the flow of traffic. For model validation, CTT (Chicago Traffic Tracker) dataset network is chosen. The parameters assessed had ensured highest prediction accuracy of 98.87%, F1-score of 0.98, 0.971 of specificity, 0.987 of recall and 0.978 of precision to forecast the presence road traffic. The proposed model demonstrates a significant advancement in road traffic detection with an accuracy of 98.87%, surpassing existing models that achieve the highest accuracy of 94.3%. This represents a notable improvement of approximately 4.77% over the highest accuracy achieved by the previous models. [ABSTRACT FROM AUTHOR]

ABSTRACT To the national economy, increasing the forecasting accuracy of realised volatility (RV) on crude oil futures markets is of critical strategic importance. However, the RV of crude oil futures cannot be accurately predicted with a single model. For this study, we adopt a hybrid model which combines gated recurrent unit (GRU) and complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) to forecast the RV of crude oil futures. Moreover, back propagation neural networks (BP), Elman neural networks (Elman), support vector regression machine (SVR), autoregressive model (AR), heterogeneous autoregressive model (HAR), and their hybrid models with CEEMDAN are adopted as comparisons. In general, this article demonstrates the superiority of the CEEMDAN‐GRU model in RV forecasting from several aspects: for both evaluation criteria, CEEMDAN‐GRU achieves the highest RV forecasting accuracy in emerging and developed crude oil futures markets; furthermore, the empirical results are robust to alternative realised measures and training sets of different lengths. [ABSTRACT FROM AUTHOR]

The number of electric vehicle (EV) in residential areas is growing rapidly, resulting in large‐scale charging of EVs connected to the distribution network. This poses a challenge to the safe and stable operation of the distribution network. In order to cope with this challenge, it is crucial to achieve accurate EV charging load prediction. However, current researches on EV charging load prediction suffer from insufficient data feature mining and lower prediction accuracy. To address this issue, this paper proposes a SE‐GRU‐MA residential EV charging load prediction method that incorporates multi‐source data feature mining. The proposed method employs a multi‐source data feature mining approach based on Pearson correlation analysis, which enhances the training efficiency and prediction accuracy of the prediction model. Additionally, this study develops a prediction model based on SE‐GRU‐MA hybrid network to achieve accurate EV charging load prediction. Simulation cases on actual history data validate that the proposed feature mining method can effectively promote prediction accuracy, and proposed SE‐GRU‐MA prediction model exhibits superior prediction capability in comparison to existing models. © 2024 Institute of Electrical Engineers of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

The main causes of aviation accidents in recent years are mostly related to pilot operational errors and pilot operational characteristics directly reflect flight quality. Hence, flight quality and flight safety are inseparable. Improving the assessment method of flight quality is of great significance for building a competency-based and evidence-based flight training system as well as enhancing flight safety. However, the problem is that some of the existing research is one-sided, and the assessment accuracy is not high. We propose a flight quality assessment method based on KOA-CNN-GRU-self-attention for the whole flight phase to accurately assess the flight quality and to improve and supplement the existing system. Firstly, the QAR data of the whole flight phase is selected and divided into three data sets according to the three indexes of operational smoothness, accuracy, and promptness, which are respectively substituted into the PCA comprehensive evaluation model to assess flight quality. Then, the evaluation results are labelled with the rating as the input of CNN-GRU-self-attention, and the parameters are optimized using KOA. Finally, the evaluation of flight quality for the three indexes was achieved by training the KOA-CNN-GRU-self-attention model. The test results show that the accuracy of operational smoothness, accuracy, and promptness reaches 98.73%, 95.07%, and 97.18%, respectively, and the assessment outcome is better and higher than the existing model. The model is also compared and analyzed with three base models CNN, QDA, XGBoost, and three fusion models CNN-self-attention, GRU-self-attention, CNN-GRU-self-attention, which show overall better results in accuracy, recall, precision, and F1-Score. [ABSTRACT FROM AUTHOR]

This paper introduces a novel deep learning model for ECG signal classification using feature fusion. The proposed methodology transforms the ECG time series into a spectrogram image using a short-time Fourier transform (STFT). This spectrogram is further processed to generate a histogram of oriented gradients (HOG) and local binary pattern (LBP) features. Three separate 2D convolutional neural networks (CNNs) then analyze these three image representations in parallel. To enhance performance, the extracted features are concatenated before feeding them into a gated recurrent unit (GRU) model. The proposed approach is extensively evaluated on two ECG datasets (MIT-BIH + BIDMC and MIT-BIH) with three and five classes, respectively. The experimental results demonstrate that the proposed approach achieves superior classification accuracy compared to existing algorithms in the literature. This suggests that the model has the potential to be a valuable tool for accurate ECG signal classification, aiding in the diagnosis and treatment of various cardiovascular disorders. [ABSTRACT FROM AUTHOR]

Spam messages have emerged as a significant issue in digital communication, adversely affecting users' mental health, personal safety, and network resources. Traditional spam detection methods often suffer from low detection rates and high false positives, underscoring the need for more effective solutions. This paper proposes the EGMA model, an ensemble learning-based hybrid approach for spam detection in SMS messages, which integrates gated recurrent unit (GRU), multilayer perceptron (MLP), and hybrid autoencoder models utilizing a majority voting algorithm. The EGMA model enhances performance by incorporating additional statistical features extracted from message content and employing text vectorization techniques, such as Term Frequency–Inverse Document Frequency (TF-IDF) and CountVectorizer. The proposed model achieved impressive classification accuracies of 99.28% on the SMS Spam Collection dataset, 99.24% on the Email Spam dataset, 99.00% on the Enron-Spam dataset, 98.71% on the Super SMS dataset, and 95.09% on UtkMl's Twitter Spam dataset. These results demonstrate that the EGMA model outperforms individual models and existing methods in the literature, providing a robust solution for enhancing spam detection performance and effectively mitigating the threats that spam messages pose in digital communication. [ABSTRACT FROM AUTHOR]

Phishing websites pose a significant threat to online security, necessitating robust detection mechanisms to safeguard users' sensitive information. This study explores the efficacy of various deep learning architectures for phishing website detection. Initially, traditional sequential models, including recurrent neural networks (RNN), long short-term memory (LSTM), and gated recurrent unit (GRU), achieve accuracies of 95%, 96%, and 96.5%, respectively, on a curated dataset. Building upon these results, hybrid architectures that combine the strengths of traditional sequential models with state-of-the-art language representation models, bidirectional encoder representations from transformers (BERT) and XLNet, are investigated. Combinations such as RNN with BERT, BERT with LSTM, BERT with GRU, RNN with XLNet, XLNet with LSTM, and XLNet with GRU are evaluated. Through experimentation, accuracies of 94.5%, 96.5%, 96.1%, 95.7%, 97.4%, and 97%, respectively, are achieved, demonstrating the effectiveness of hybrid deep learning architectures in enhancing phishing detection performance. These findings contribute to advancing the state-of-the-art in cybersecurity practices and underscore the importance of leveraging diverse model types to combat online threats effectively. [ABSTRACT FROM AUTHOR]

In the domain of smart health devices, the accurate detection of physical indicators levels plays a crucial role in enhancing safety and well-being. This paper introduces a cross device federated learning framework using hybrid deep learning model. Specifically, the paper presents a comprehensive comparison of different combination of long short-term memory (LSTM), gated recurrent unit (GRU), convolutional neural network (CNN), random forest (RF), and extreme gradient boosting (XGBoost), in order to forecast stress levels by utilizing time series information derived from wearable smart gadgets. The LSTM-RF model demonstrated the highest level of accuracy, achieving 93.53% for user 1, 99.40% for user 2, and 97.88% for user 3. Similarly, the LSTM-XGBoost model yielded favorable outcomes, with accuracy rates of 85.88%, 98.55%, and 92.02% for users 1, 2, and 3, respectively, out of 23 users studied. These findings highlight the efficacy of federated learning and the utilization of hybrid models in stress detection. Unlike traditional centralized learning paradigms, the presented federated approach ensures privacy preservation and reduces data transmission requirements by processing data locally on Edge devices. [ABSTRACT FROM AUTHOR]

In the realm of artificial intelligence and musicology, emotion recognition in music performance has emerged as a pivotal area of research. This paper introduces EmoTrackNet, an integrated deep learning framework that combines sparse attention networks, enhanced one-dimensional residual Convolutional Neural Networks (CNNs) with an improved Inception module, and Gate Recurrent Units (GRUs). The synergy of these technologies aims to decode complex emotional cues embedded in music. Our methodology revolves around leveraging the sparse attention network to efficiently process temporal sequences, thereby capturing the intricate dynamics of musical pieces. The incorporation of the 1D residual CNN with an upgraded Inception module facilitates the extraction of nuanced features from audio signals, encompassing a broad spectrum of musical tones and textures. The GRU component further refines the model's capability to retain and process sequential information over longer timeframes, essential for understanding evolving emotional expressions in music. We evaluated EmoTrackNet on the Soundtrack dataset a comprehensive collection of music pieces annotated with emotional labels. The results demonstrated remarkable improvements in the accuracy of emotion recognition, outperforming existing models. This enhanced performance can be attributed to the integrated approach, which efficiently combines the strengths of each component, leading to a more robust and sensitive emotion detection system. EmoTrackNet's novel architecture and promising results pave the way for new avenues in musicology, particularly in understanding and interpreting the emotional depth of musical performances. This framework not only contributes significantly to the field of music emotion recognition but also has potential applications in music therapy, entertainment, and interactive media where emotional engagement is key. [ABSTRACT FROM AUTHOR]

To address the challenge of constructing spatial and temporal dependencies in existing traffic flow prediction methods, this paper proposed a new method called spatial temporal position attention graph neural network (ST-PAGNN), which utilized spatiotemporal location attention. Firstly, the graph neural network contained a location attention mechanism, which could better capture the spatial dependence of traffic nodes in the urban road network. Then, it used a gated recurrent neural network with trend adaptive transformer (Trendformer) to capture the local and global information of the traffic flow sequence in the time dimension. Finally, it used the improved grid search optimization method to optimize the introduced para-meters of the model, obtaining the global optimal solution with high time efficiency. The experimental results show that in the dataset PEMS-BAY, the evaluation indexes RMSE, MAE and MAPE of the ST-PAGNN method are 1.37, 2.57, 2.67%, 1.55, 3.64, 3.37%, 1.97, 4. 37 and 4.43%, respectively, when the prediction step size is 15 min, 30 min and 60 min, respectively. In the dataset METR-LA, when the prediction step size is 15 min, 30 min and 60 min, the evaluation indexes RMSE, MAE and MAPE of the ST-PAGNN method are 2. 73, 5. 16, 7.13%, 2.99, 5.97, 7.86%, 3.53, 7. 16 and 9.96%, respectively. The results show that the proposed ST-PAGNN method is higher than the existing models in the evaluation indexes under different granularities, which illustrates the effectiveness and superiority of ST-PAGNN in solving traffic prediction problems. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Intelligent Transportation Systems & Applications is the property of Journal of Intelligent Transportation Systems & Applications and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Credit card fraud detection is a critical challenge in the financial industry, with substantial economic implications. Conventional machine learning (ML) techniques often fail to adapt to evolving fraud patterns and underperform with imbalanced datasets. This study proposes a hybrid deep learning framework that integrates Generative Adversarial Networks (GANs) with Recurrent Neural Networks (RNNs) to enhance fraud detection capabilities. The GAN component generates realistic synthetic fraudulent transactions, addressing data imbalance and enhancing the training set. The discriminator, implemented using various DL architectures, including Simple RNN, Long Short-Term Memory (LSTM) networks, and Gated Recurrent Units (GRUs), is trained to distinguish between real and synthetic transactions and further fine-tuned to classify transactions as fraudulent or legitimate. Experimental results demonstrate significant improvements over traditional methods, with the GAN-GRU model achieving a sensitivity of 0.992 and specificity of 1.000 on the European credit card dataset. This work highlights the potential of GANs combined with deep learning architectures to provide a more effective and adaptable solution for credit card fraud detection. [ABSTRACT FROM AUTHOR]

Time series prediction plays a crucial role in fields such as finance, economy, weather, and energy. Traditionally, the widely used SARIMAX model considers the seasonality, trend, and exogenous variables of time series data but has limitations in capturing nonlinear patterns due to linear assumptions. In contrast, recurrent neural network models based on deep learning, especially GRU, are attracting attention for their ability to effectively model nonlinearity. This study aims to improve time series prediction accuracy by proposing a Stacked SARIMAX-GRU model that combines the linear prediction ability of the SARIMAX model with the nonlinear prediction ability of the GRU model. This model enhances prediction accuracy by using the results of the two models as input to the metamodel through the stacking technique. Empirical analysis confirmed that the proposed model showed better prediction performance than using the existing SARIMAX and GRU models alone, with reduced errors such as MAE, MAPE, and RMSE, and a higher R-squared value. [ABSTRACT FROM AUTHOR]

The widespread use of technology has led to a transformation of human behaviors and habits into the digital space; and generating extensive data plays a crucial role when coupled with forecasting techniques in guiding marketing decision-makers and shaping strategic choices. Traditional methods like autoregressive moving average (ARMA) can-not be used at predicting individual behaviors because we can-not create models for each individual and buy till you die (BTYD) models have limitations in capturing the trends accurately. Recognizing the paramount importance of individual-level predictions, this study proposes a deep learning framework, specifically uses gated recurrent unit (GRU), for enhanced behavior analysis. This article discusses the performance of GRU and long short-term memory (LSTM) models in this framework for forecasting future individual behaviors and presenting a comparative analysis against benchmark BTYD models. GRU and LSTM yielded the best results in capturing the trends, with GRU demonstrating a slightly superior performance compared to LSTM. However, there is still significant room for improvement at the individual level. The findings not only demonstrate the performance of GRU and LSTM models but also provide valuable insights into the potential of new techniques or approaches for understanding and predicting individual behaviors. [ABSTRACT FROM AUTHOR]

Abnormal signals of brain activity can predict epilepsy, which can be effectively detected with the use of IoT-enabled Electro-Encephalo-Gram (EEG) devices. In this process, wearable devices can collect relevant data and transmit them to health providers for analysis. These data can be assessed for epilepsy using Deep Learning (DL) algorithms. DL and evolutionary algorithms are combined to detect epilepsy detection with optimized performance. This study proposed a system with multiple objectives. First, EEG signals were obtained using IoT from subjects in healthy conditions and with epilepsy. In preprocessing, the EEG signal is filtered using finite impulse response. Features were extracted from preprocessed signals, including wavelet coefficients, signal entropy, spectral power, coherence, and frequency bands. An optimal structure was selected from the extracted features through a newly designed hybrid optimization model, called the alpha bat customized squirrel optimizer, with a combination of standard jellyfish search algorithm with particle swarm optimization. Finally, a multimodal deep learning framework, including Long Short-Term Memory Network (LSTM), Gated Recurrent Units (GRU), and Convolutional Neural Network (CNN), detects epilepsy. The results show that the proposed multilayer DL-based approach outperforms existing methods in terms of accuracy, precision, sensitivity, False Negative Rate (FNR), and specificity. [ABSTRACT FROM AUTHOR]

Maritime transportation, a cornerstone of global trade, faces increasing safety challenges due to growing sea traffic volumes. This study proposes a novel approach to vessel trajectory prediction utilizing Automatic Identification System (AIS) data and advanced deep learning models, including Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Bidirectional LSTM (DBLSTM), Simple Recurrent Neural Network (SimpleRNN), and Kalman Filtering. The research implemented rigorous AIS data preprocessing, encompassing record deduplication, noise elimination, stationary simplification, and removal of insignificant trajectories. Models were trained using key navigational parameters: latitude, longitude, speed, and heading. Spatiotemporal aware processing through trajectory segmentation and topological data analysis (TDA) was employed to capture dynamic patterns. Validation using a three-month AIS dataset demonstrated significant improvements in prediction accuracy. The GRU model exhibited superior performance, achieving training losses of 0.0020 (Mean Squared Error, MSE) and 0.0334 (Mean Absolute Error, MAE), with validation losses of 0.0708 (MSE) and 0.1720 (MAE). The LSTM model showed comparable efficacy, with training losses of 0.0011 (MSE) and 0.0258 (MAE), and validation losses of 0.2290 (MSE) and 0.2652 (MAE). Both models demonstrated reductions in training and validation losses, measured by MAE, MSE, Average Displacement Error (ADE), and Final Displacement Error (FDE). This research underscores the potential of advanced deep learning models in enhancing maritime safety through more accurate trajectory predictions, contributing significantly to the development of robust, intelligent navigation systems for the maritime industry. [ABSTRACT FROM AUTHOR]

Copyright of Machine Tool & Hydraulics is the property of Guangzhou Mechanical Engineering Research Institute (GMERI) and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The Internet of Things' (IoT) rapid growth has resulted in a rise in vulnerabilities, making safeguarding IoT systems against intrusions and illegal access a top priority. Intrusion Detection Systems (IDS) are essential for keeping an eye out for irregularities in network traffic. However, the challenge lies in the IDS's ability to detect attacks within high-speed networks while minimizing computational complexity promptly. To improve detection efficiency in IoT networks, we proposed lightweight detection models in this paper that are based on Long Short-Term Memory (LSTM), Bidirectional LSTM (Bi-LSTM), Gated Recurrent Unit (GRU), and a GRU-based self-attention mechanism. The Grid Search (GS) algorithm optimizes the models by adjusting the hyperparameters, such as the learning rate and the number of hidden units. The proposed models are evaluated using the ToN-IoT dataset. The achieved detection accuracy for all models is as follows: 97% for GRU, 98.1% for LSTM, 98.4% for Bi-LSTM, and 99% for the GRU-based self-attention mechanism. Furthermore, the GRU-based self-attention mechanism has fewer parameters, which leads to a significant saving in classification time of up to 84% compared to GRU. These findings demonstrate that the GRU-based self-attention mechanism is superior in accuracy and computational efficiency, which makes it particularly effective for real-time intrusion detection in IoT networks. [ABSTRACT FROM AUTHOR]

Code smells indicate potential symptoms or problems in software due to inefficient design or incomplete implementation. These problems can affect software quality in the long-term. Code smell detection is fundamental to improving software quality and maintainability, reducing software failure risk, and helping to refactor the code. Previous works have applied several prediction methods for code smell detection. However, many of them show that machine learning (ML) and deep learning (DL) techniques are not always suitable for code smell detection due to the problem of imbalanced data. So, data imbalance is the main challenge for ML and DL techniques in detecting code smells. To overcome these challenges, this study aims to present a method for detecting code smell based on DL algorithms (Bidirectional Long Short-Term Memory (Bi-LSTM) and Gated Recurrent Unit (GRU)) combined with data balancing techniques (random oversampling and Tomek links) to mitigate data imbalance issue. To establish the effectiveness of the proposed models, the experiments were conducted on four code smells datasets (God class, data Class, feature envy, and long method) extracted from 74 open-source systems. We compare and evaluate the performance of the models according to seven different performance measures accuracy, precision, recall, f-measure, Matthew's correlation coefficient (MCC), the area under a receiver operating characteristic curve (AUC), the area under the precision–recall curve (AUCPR) and mean square error (MSE). After comparing the results obtained by the proposed models on the original and balanced data sets, we found out that the best accuracy of 98% was obtained for the Long method by using both models (Bi-LSTM and GRU) on the original datasets, the best accuracy of 100% was obtained for the long method by using both models (Bi-LSTM and GRU) on the balanced datasets (using random oversampling), and the best accuracy 99% was obtained for the long method by using Bi-LSTM model and 99% was obtained for the data class and Feature envy by using GRU model on the balanced datasets (using Tomek links). The results indicate that the use of data balancing techniques had a positive effect on the predictive accuracy of the models presented. The results show that the proposed models can detect the code smells more accurately and effectively. [ABSTRACT FROM AUTHOR]

The Non-orthogonal multiple access (NOMA) technique has drawn considerable attention as a promising solution for advanced wireless communication systems due to its higher data rates, lower latency, and high bandwidth efficiency. The decoding of NOMA signals requires standard successive interference cancellation (SIC) approaches. Due to the propagation delay and fading channels, multipath fading significantly impacts the SIC process and correct signal detection. This article proposes a hybrid approach using deep convolutional neural network-gated recurrent unit-support vector machine (CNN-GRU (CGRU)-SVM) to detect the uplink NOMA users' signals. The proposed deep learning (DL) CGRU-SVM-based receiver executes simultaneous channel estimation, equalization, and demodulation, performing perfect end-to-end operations. The simulation results discuss the bit error rate performance of the proposed CGRU-SVM-based receiver with the conventional least square and minimum mean-square error signal detection methods, along with other DL-based GRU, CNN, and CGRU approaches in uplink NOMA schemes. It provides the efficient extraction of spatiotemporal features with an SVM classifier. Moreover, the proposed model outperforms the other DL-based techniques concerning the accuracy, F1 score, and receiver operating characteristic with the area under the curve. [ABSTRACT FROM AUTHOR]

Stock market investments have become an essential part of our lives as they offer a means of growing wealth and securing financial stability for individuals and businesses alike. However, predicting and investing in a stock is a complex method and demands significant levels of knowledge, proficiency, and skill. Stock market prediction is the act of analyzing historical information and trends in the market in order to make informed forecasts about the potential future worth of a given stock or instrument that is traded on a financial instrument exchange. When making stock predictions, most stockbrokers use both fundamental and technical analysis and time series analysis. The study focuses on the implementation of Multi-Linear Regression, LSTM, CNN, and LSTM + GRU based Machine learning techniques using technical analysis to predict stock's closing values from the NYSE, and NASDAQ markets for multiple days. The dataset has been taken from Yahoo, of 10-year span. The factors taken into consideration for predicting stock prices are open, close, low, and high. The model's effectiveness is measured using common strategic metrics like RMSE, MSE, MAE, and R2. A lower value for these variables suggests that the models are good at forecasting stock closing prices. After conducting a comprehensive evaluation, we found that LSTM + GRU model performs the best among the tested models for predicting multiple days, followed by CNN and LSTM. The tested models demonstrate a remarkable level of accuracy in predicting stock market prices. This research work provides a valuable contribution to the fields of financial and technical analysis in the stock market research community. [ABSTRACT FROM AUTHOR]

Recurrent neural networks (RNNs) have significantly advanced the field of machine learning (ML) by enabling the effective processing of sequential data. This paper provides a comprehensive review of RNNs and their applications, highlighting advancements in architectures, such as long short-term memory (LSTM) networks, gated recurrent units (GRUs), bidirectional LSTM (BiLSTM), echo state networks (ESNs), peephole LSTM, and stacked LSTM. The study examines the application of RNNs to different domains, including natural language processing (NLP), speech recognition, time series forecasting, autonomous vehicles, and anomaly detection. Additionally, the study discusses recent innovations, such as the integration of attention mechanisms and the development of hybrid models that combine RNNs with convolutional neural networks (CNNs) and transformer architectures. This review aims to provide ML researchers and practitioners with a comprehensive overview of the current state and future directions of RNN research. [ABSTRACT FROM AUTHOR]

Abnormal behavior recognition is an important technology used to detect and identify activities or events that deviate from normal behavior patterns. It has wide applications in various fields such as network security, financial fraud detection, and video surveillance. In recent years, Deep Convolution Networks (ConvNets) have been widely applied in abnormal behavior recognition algorithms and have achieved significant results. However, existing abnormal behavior detection algorithms mainly focus on improving the accuracy of the algorithms and have not explored the real-time nature of abnormal behavior recognition. This is crucial to quickly identify abnormal behavior in public places and improve urban public safety. Therefore, this paper proposes an abnormal behavior recognition algorithm based on three-dimensional (3D) dense connections. The proposed algorithm uses a multi-instance learning strategy to classify various types of abnormal behaviors, and employs dense connection modules and soft-threshold attention mechanisms to reduce the model's parameter count and enhance network computational efficiency. Finally, redundant information in the sequence is reduced by attention allocation to mitigate its negative impact on recognition results. Experimental verification shows that our method achieves a recognition accuracy of 95.61% on the UCF-crime dataset. Comparative experiments demonstrate that our model has strong performance in terms of recognition accuracy and speed. [ABSTRACT FROM AUTHOR]

Background: Arabic image captioning (AIC) is the automatic generation of text descriptions in the Arabic language for images. Applies a transfer learning approach in deep learning to enhance computer vision and natural language processing. There are many datasets in English reverse other languages. Instead of, the Arabs researchers unanimously agreed that there is a lack of Arabic databases available in this field. Objective: This paper presents the improvement and processing of the available Arabic textual database using Google spreadsheets for translation and creation of AR. Flicker8k2023 dataset is an extension of the Arabic Flicker8k dataset available, it was uploaded to GitHub and made public for researches. Methods: An efficient model proposed using deep learning techniques by including two pre-training models (VGG16 and VGG19), to extract features from the images and build (LSTM and GRU) models to process textual prediction sequence. In addition to the effect of pre-processing the text in Arabic. Results: The adopted model outperforms better compared to the previous study in BLEU-1 from 33 to 40. Conclusions: This paper concluded that the biggest problem is the database available in the Arabic language. This paper has worked to increase the size of the text database from 24,276 to 32,364 thousand captions, where each image contains 4 captions. [ABSTRACT FROM AUTHOR]

Speech recognition is a critical task in spoken language applications. Globally known models such as DeepSpeech2 are effective for English speech recognition, however, they are not well-suited for languages like Arabic. This paper is interested in recognizing the Arabic language, especially Modern Standard Arabic (MSA). This paper proposed two models that utilize "Franco-Arabic" as an encoding mechanism and additional enhancements to recognize MSA. The first model uses Mel-Frequency Cepstral Coefficients (MFCCs) as input features, while the second employs six sequential Gated Recurrent Unit (GRU) layers. Each model is then followed by a fully connected layer with a dropout layer which helped reduce overfitting. The Connectionist Temporal Classification (CTC) loss is used to calculate the prediction error and to maximize the correct transcription likelihood. Two experiments were conducted for each model. The first experiment involved 41 h of continuous speech over 15 epochs. Whereas, the second one utilized 69 h over 30 epochs. The experiments showed that the first model excels in speed while the second excels in accuracy, and both outperformed the well-known DeepSpeech2. [ABSTRACT FROM AUTHOR]

This study addresses the integration of machine learning (ML) with supervisory control and data acquisition (SCADA) systems to enhance predictive maintenance and operational efficiency in oil well monitoring. We investigated the applicability of advanced ML models, including Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), and Momentum LSTM (MLSTM), on a dataset of 21,644 operational records. These models were trained to predict a critical operational parameter, FlowRate, which is essential for operational integrity and efficiency. Our results demonstrate substantial improvements in predictive accuracy: the LSTM model achieved an R2 score of 0.9720, the BiLSTM model reached 0.9725, and the MLSTM model topped at 0.9726, all with exceptionally low Mean Absolute Errors (MAEs) around 0.0090 for LSTM and 0.0089 for BiLSTM and MLSTM. These high R2 values indicate that our models can explain over 97% of the variance in the dataset, reflecting significant predictive accuracy. Such performance underscores the potential of integrating ML with SCADA systems for real-time applications in the oil and gas industry. This study quantifies ML's integration benefits and sets the stage for further advancements in autonomous well-monitoring systems. [ABSTRACT FROM AUTHOR]

In recent years, deep learning (DL) methods, such as recurrent neural networks (RNN). have been used for streamflow prediction. In this study, the monthly inflow into the Dez Dam reservoir from 1955 to 2018 in southwestern Iran was simulated using various types of RNNs, including long short-term memory (LSTM), bidirectional long short-term memory (Bi-LSTM), gated recurrent unit (GRU), and stacked long short-term memory (Stacked LSTM). It was observed that considering flow discharge, temperature, and precipitation as inputs to the models yields the best results. Additionally, wavelet transform was employed to enhance the accuracy of the RNNs. Among the RNNs, the GRU model exhibited the best performance in simulating monthly streamflow without using wavelet transform, with RMSE, MAE, NSE, and R2 values of 0.061 m3/s, 0.038 m3/s, 0.556, and 0.642, respectively. Moreover, in the case of using wavelet transform, the Bi-LSTM model with db5 mother wavelet and decomposition level 5 was able to simulate the monthly streamflow with high accuracy, yielding RMSE, MAE, NSE, and R2 values of 0.014 m3/s, 0.008 m3/s, 0.9983, and 0.9981, respectively. Uncertainty analysis was conducted for the two mentioned superior models. To quantify the uncertainty, the concept of the 95 percent prediction uncertainty (95PPU) and the p-factor and r-factor criteria were utilized. For the GRU, the p-factor and r-factor values were 82% and 1.28, respectively. For the Bi-LSTM model, the p-factor and r-factor values were 94% and 1.06, respectively. The obtained p-factor and r-factor values for both models are within the acceptable and reliable range. [ABSTRACT FROM AUTHOR]

The malfunction variables of power stations are related to the areas of weather, physical structure, control, and load behavior. To predict temporal power failure is difficult due to their unpredictable characteristics. As high accuracy is normally required, the estimation of failures of short-term temporal prediction is highly difficult. This study presents a method for converting stochastic behavior into a stable pattern, which can subsequently be used in a short-term estimator. For this conversion, K-means clustering is employed, followed by long-short-term memory and gated recurrent unit algorithms are used to perform the short-term estimation. The environment, the operation, and the generated signal factors are all simulated using mathematical models. Weather parameters and load samples have been collected as part of a dataset. Monte-Carlo simulation using MATLAB programming has been used to conduct experimental estimation of failures. The estimated failures of the experiment are then compared with the actual system temporal failures and found to be in good match. Therefore, to address the gap in knowledge for any future power grid estimated failures, the achieved results in this paper form good basis for a testbed to estimate any grid future failures. [ABSTRACT FROM AUTHOR]

Deoxyribonucleic acid (DNA) profiling is an important branch of forensic science that aids in the identification of missing people, particularly in mass disasters. This study presents an artificial intelligence system that utilizes DNA-Short Tandem Repeat (STR) data to identify victims using Deep Neural Network (DNN), Gated Recurrent Unit (GRU), and Bidirectional GRU (Bi-GRU) deep learning models. The identification of STR information for living family members, such as parents or brothers, poses a significant challenge in victim identification. Familial data are artificially generated based on the actual data of distinct Iraqi individuals from the province of Al-Najaf. Two people are selected as male and female to create a family of 10 members. As a result of this action, 151,580 individuals were generated from 106 different people, which helps to overcome the lack of datasets caused by restrictive policies and the confidentiality of familial datasets in Iraq. These datasets are prepared and formatted for training deep learning models. Based on various reference datasets, the models are built to handle five different scenarios where both parents are alive, only one parent is alive, or the siblings are available for reference. The three models' performances were compared: Bi-GRU performed the best, with a loss of 0.0063 and an accuracy of 0.9979, followed by GRU with a loss of 0.0102 and an accuracy of 0.9964, and DNN with a loss of 0.2276 and an accuracy of 0.9174. The evaluation makes use of a confusion matrix and receiver operating characteristic curve. Based on the literature, this is the first attempt to introduce deep learning in DNA profiling, which reduces both time and effort. Despite the fact that the proposed deep learning models have good results in identifying missing persons according to their families, these models have limitations that can be confined to the availability of familial DNA profiles. The system doesn't work well if no relative samples are available as references, such as a father, mother, or brother. In the future, DNN, GRU, and Bi-GRU models will be applied to mini-STR sequences that are used in cases of degraded victims of incomplete STR sequences. [ABSTRACT FROM AUTHOR]

The foreign exchange market is the most liquid financial market globally, attracting investors looking for lucrative investment opportunities. Despite numerous techniques developed for forecasting foreign exchange trends, accurate and reliable models remain scarce. This article presents a novel approach that combines fundamental and technical analysis to predict exchange rates for the USD-CNY, EUR-USD, and GBP-USD currency pairs. Additionally, we extend the model's architecture by using China CSI300 stock index futures (CIFc1) instead of VIX, LSTM instead of GRU, and adding data pre-processing. The results show that our method is more accurate and stable than other approaches mentioned above, including traditional methods based on fundamental analysis. This study highlights the importance of the idea of combing fundamental information with deep learning, and underscores the effectiveness of integrating technique analysis and fundamental analysis, and lays the groundwork for further extensions and experimentation in foreign exchange forecasting. [ABSTRACT FROM AUTHOR]

The management of Municipal Solid Waste (MSW) establishes a decisive facet in the development of sustainable and vigorous societies that requires an adequate framework in order to prevent a risk of environmental pollution. Therefore, it's imperative to establish a robust framework to handle the generated MSW. In this regard, early predictions of waste generation with high accuracy emerges as a pivotal factor in serving the municipal authorities to formulate an effective MSW management system. Various researchers have significantly contributed to the field of early waste predictions by developing different deep learning models to attain the highly accurate predicted results. However, it is important to note that each model has its strengths and limitations. This study has predominantly focused on the enhancement of the Gated Recurrent Unit (GRU) model. The primary limitation in the standard GRU model lies in both of its gates (reset and update) processing the identical data. This redundancy has contributed to a prolonged training time and a diminished convergence rate. Therefore, this study has proposed an enhanced GRU model with optimized update gate (EGRU-OU) to address the redundancy issue that lies between the two gates. The EGRU-OU model will provide the filtered data specifically to the update gate which is instrumental in significantly reducing the redundant information between the two gates. Moreover, this study has employed two different datasets to conduct a comprehensive analysis of the results. One dataset has been collected for 16 different developed countries, whereas, the other dataset has been collected for Multan City, Pakistan. These datasets have been segmented into three subdivisions: training data, constituting 70% of the dataset for model training; testing data, comprising 15% for model evaluation; and validation data, representing the remaining 15% for additional model validation. In addition, The EGRU-OU model has been compared with other benchmark models including standard GRU model, Long Short Term Memory (LSTM) and Artificial Neural Network (ANN) model based on three distinct error metrics: Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the coefficient of determination (R²). The outcomes have clearly demonstrated a superior performance of the EGRU-OU approach as compared to other models with the least error matrices values of MAE being 0.036 and RMSE being 0.0684. [ABSTRACT FROM AUTHOR]

Power consumption in the home has grown in recent years as a consequence of the use of varied residential applications. On the other hand, many families are beginning to use renewable energy, such as energy production, energy storage devices, and electric vehicles. As a result, estimating household power demand is necessary for energy consumption monitoring and planning. Power consumption forecasting is a challenging time series prediction topic. Furthermore, conventional forecasting approaches make it difficult to anticipate electric power consumption since it comprises irregular trend components, such as regular seasonal fluctuations. To address this issue, algorithms combining stationary wavelet transform (SWT) with deep learning models have been proposed. The denoised series is fitted with various benchmark models, including Long Short-Term Memory (LSTM), Gated Recurrent Units (GRUs), Bidirectional Gated Recurrent Units (Bi-GRUs), Bidirectional Long Short-Term Memory (Bi-LSTM), and Bidirectional Gated Recurrent Units Long Short-Term Memory (Bi-GRU LSTM) models. The performance of the SWT approach is evaluated using power consumption data at three different time intervals (1 min, 15 min, and 1 h). The performance of these models is evaluated using metrics such as Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and Mean Absolute Error (MAE). The SWT/GRU model, utilizing the bior2.4 filter at level 1, has emerged as a highly reliable option for precise power consumption forecasting across various time intervals. It is observed that the bior2.4/GRU model has enhanced accuracy by over 60% compared to the deep learning model alone across all accuracy measures. The findings clearly highlight the success of the SWT denoising technique with the bior2.4 filter in improving the power consumption prediction accuracy. [ABSTRACT FROM AUTHOR]

Autism spectrum disorder (ASD) is a complex, severe disorder related to brain development. It impairs patient language communication and social behaviors. In recent years, ASD researches have focused on a single-modal neuroimaging data, neglecting the complementarity between multi-modal data. This omission may lead to poor classification. Therefore, it is important to study multi-modal data of ASD for revealing its pathogenesis. Furthermore, recurrent neural network (RNN) and gated recurrent unit (GRU) are effective for sequence data processing. In this paper, we introduce a novel framework for a Multi-Kernel Learning Fusion algorithm based on RNN and GRU (MKLF-RAG). The framework utilizes RNN and GRU to provide feature selection for data of different modalities. Then these features are fused by MKLF algorithm to detect the pathological mechanisms of ASD and extract the most relevant the Regions of Interest (ROIs) for the disease. The MKLF-RAG proposed in this paper has been tested in a variety of experiments with the Autism Brain Imaging Data Exchange (ABIDE) database. Experimental findings indicate that our framework notably enhances the classification accuracy for ASD. Compared with other methods, MKLF-RAG demonstrates superior efficacy across multiple evaluation metrics and could provide valuable insights into the early diagnosis of ASD. [ABSTRACT FROM AUTHOR]

Power consumption in the home has grown in recent years as a consequence of the use of varied residential applications. On the other hand, many families are beginning to use renewable energy, such as energy production, energy storage devices, and electric vehicles. As a result, estimating household power demand is necessary for energy consumption monitoring and planning. Power consumption forecasting is a challenging time series prediction topic. Furthermore, conventional forecasting approaches make it difficult to anticipate electric power consumption since it comprises irregular trend components, such as regular seasonal fluctuations. To address this issue, algorithms combining stationary wavelet transform (SWT) with deep learning models have been proposed. The denoised series is fitted with various benchmark models, including Long Short-Term Memory (LSTM), Gated Recurrent Units (GRUs), Bidirectional Gated Recurrent Units (Bi-GRUs), Bidirectional Long Short-Term Memory (Bi-LSTM), and Bidirectional Gated Recurrent Units Long Short-Term Memory (Bi-GRU LSTM) models. The performance of the SWT approach is evaluated using power consumption data at three different time intervals (1 min, 15 min, and 1 h). The performance of these models is evaluated using metrics such as Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and Mean Absolute Error (MAE). The SWT/GRU model, utilizing the bior2.4 filter at level 1, has emerged as a highly reliable option for precise power consumption forecasting across various time intervals. It is observed that the bior2.4/GRU model has enhanced accuracy by over 60% compared to the deep learning model alone across all accuracy measures. The findings clearly highlight the success of the SWT denoising technique with the bior2.4 filter in improving the power consumption prediction accuracy.

Abstract To quickly assess slope stability based on field displacement monitoring data, this paper constructs a hybrid optimization model that predicts surface displacement during tunnel excavation in base-overburden slopes. The model combines Wavelet Decomposition (WD) with a Gated Recurrent Unit (GRU), and the GRU's hyperparameters are optimized using an Improved Particle Swarm Optimization algorithm (IPSO). The specific steps are as follows: First, the Wavelet Decomposition (WD) technique is applied to decompose the raw displacement data, extracting features at different time–frequency scales. Next, the Dropout technique is incorporated into the GRU model to prevent overfitting. Additionally, nonlinear inertia weight ω improved cognitive factor c 1 , and social factor c 2 are introduced. The PSO algorithm is improved by integrating crossover and mutation concepts from genetic algorithms. Finally, the IPSO is used to optimize the number of neural units h N , H N , L N and dropout rates D 1 and D 2 in the GRU network architecture. After constructing the WD-IPSO-GRU model, a comprehensive comparison is made with various swarm intelligence algorithms and state-of-the-art models. The experimental results demonstrate that the WD-IPSO-GRU model significantly improves the prediction accuracy of surface displacement in slopes during tunnel excavation. Compared to directly using raw data for prediction, the introduction of the WD preprocessing technique improved the prediction accuracy at measurement points 01 and 02 by 28% and 45.9%, respectively. Additionally, with the model optimized by IPSO, the prediction accuracy at measurement points 01 and 02 increased by 76% and 56.7%, respectively. The WD-IPSO-GRU model effectively addresses the challenges of extracting features from univariate displacement time-series data and determining the parameters of the GRU network. It improves the prediction accuracy of surface displacement in base-overburden type slopes and demonstrates excellent generalization ability and reliability. The research results validate the potential application of the model in geotechnical engineering and provide strong support for assessing slope stability during tunnel excavation.

Forecasting energy demand is essential for efficient grid management as it promotes steady operations, efficient markets, and sustainable energy practices. In this study, previously observed, evenly spaced energy consumption data are analysed using recurrent neural networks based on Gated Recurrent Unit (GRU) and Long Short-Term Memory (LSTM) architectures to extract important insights, features, and remarkable patterns. First, the study examines the influence of meteorological features on energy consumption. The most significant meteorological features are determined by computing the MIC and Pearson's correlation coefficient. The selected features are then combined with historical energy consumption data to feed the neural network. Second, to improve and optimise the performance of the proposed models, two technical indicators - the daily energy usage average and the simple moving average - are considered. The following are some instances of comparisons in terms of prediction accuracy: (1) The MAPE of the proposed model is 2.47, whereas that of the current model is 4.03. (2) The MAPE of the existing model is 25.83, whereas the proposed solution is 18.68. (3) The MAPE of the suggested model is 24.8, while the MAPE of the current model is 26.6. (4) The MAPE of the present model is 4.77, whereas the suggested approach's is 4.42. [ABSTRACT FROM AUTHOR]

Global food security, economic growth, and biodiversity preservation are impacted significantly by aquaculture. Water quality monitoring (WQM) and water quality prediction (WQP) are essential for profitable as well as sustainable aquaculture. Empirical techniques lead to erroneous WQP, which has a negative impact on aquaculture by generating disease outbreaks, oxygen depletion, nutrient imbalances, chemical pollution, and unfavorable environmental effects. In this work, we propose attention-driven long short-term memory (A-LSTM) and gated recurrent unit (A-GRU) deep learning recurrent neural network (DL-RNN) models for aquaculture WQP. This study utilizes two datasets. The first dataset consists of 3 years of data with 1096 samples collected from aquaculture farms under the Agency for Development of Aquaculture Kerala (ADAK) in India. The second dataset is publicly available, where data is collected from the marine aquaculture base in Xincun Town, LingShui County, Hainan Province, China, which consists of 23200 samples collected over 80 days. Additionally, this work presents a thorough analysis of the effects of hyperparameters ( h p ) on the performance of the proposed models using two different water quality datasets. The prediction performance of proposed A-LSTM as well as A-GRU are compared with conventional LSTM and GRU DL-RNN models in terms of prediction accuracy and computational efficiency. Prediction accuracy in the range of 98.30 to 99.70% is observed for various water quality parameters. The findings demonstrate that the proposed A-LSTM and A-GRU models significantly improve prediction accuracy with lesser computation time. [ABSTRACT FROM AUTHOR]

Deep learning models (DLMs), such as recurrent neural networks (RNN), long short-term memory (LSTM), bidirectional long short-term memory (Bi-LSTM), and gated recurrent unit (GRU), are superior for sequential data analysis due to their ability to learn complex patterns. This paper proposes enhancing performance of these models by applying fuzzy c-means (FCM) clustering on sequential data from a nonlinear plant and the stock market. FCM clustering helps to organize the data into clusters based on similarity, which improves the performance of the models. Thus, the proposed fuzzy c-means recurrent neural network (FCM-RNN), fuzzy c-means long short-term memory (FCM-LSTM), fuzzy c-means bidirectional long short-term memory (FCM-Bi-LSTM), and fuzzy c-means gated recurrent unit (FCM-GRU) models showed enhanced prediction results than RNN, LSTM, Bi-LSTM, and GRU models, respectively. This enhancement is validated using performance metrics such as root-mean-square error and mean absolute error and is further illustrated by scatter plots comparing actual versus predicted values for training, validation, and testing data. The experiment results confirm that integrating FCM clustering with DLMs shows the superiority of the proposed models. [ABSTRACT FROM AUTHOR]

Due to the limitations of the model itself, the performance of switched autoregressive exogenous (SARX) models will face potential threats when modeling nonlinear hybrid dynamic systems. To address this problem, a robust identification approach of the switched gated recurrent unit (SGRU) model is developed in this paper. Firstly, all submodels of the SARX model are replaced by gated recurrent unit neural networks. The obtained SGRU model has stronger nonlinear fitting ability than the SARX model. Secondly, this paper departs from the conventional Gaussian distribution assumption for noise, opting instead for a generalized Gaussian distribution. This enables the proposed model to achieve stable prediction performance under the influence of different noises. Notably, no prior assumptions are imposed on the knowledge of operating modes in the proposed switched model. Therefore, the EM algorithm is used to solve the problem of parameter estimation with hidden variables in this paper. Finally, two simulation experiments are performed. By comparing the nonlinear fitting ability of the SGRU model with the SARX model and the prediction performance of the SGRU model under different noise distributions, the effectiveness of the proposed approach is verified. [ABSTRACT FROM AUTHOR]

Traditional methods of authentication, such as keys, passwords, and PIN codes, are increasingly being surpassed by the robustness and security offered by biometric systems. This study presents a system for human identification using electrocardiogram (ECG) signals, evaluated using two deep learning models: gated recurrent units (GRU) and long short-term memory (LSTM). The proposed methodology begins with signal denoising using a bandpass filter, followed by the decomposition of the signal into multiple intrinsic mode functions (IMFs) via a signal processing technique, empirical mode decomposition (EMD). The first two IMFs, after normalization and segmentation, serve as features for the deep learning models. The performance of these models was assessed on three distinct databases: NSRDB, MIT-BIH, and PTB. MIT-BIH contains a mixture of healthy and ill subjects, while the PTB and NSRDB databases contain only healthy subjects. The GRU model achieved accuracies of 98.57%, 98.26%, and 99.17% on these databases respectively, while the LSTM model achieved accuracies of 98.33%, 97.89%, and 98.27% respectively, demonstrating its reliability for biometric applications. While the proposed system was tested only on ECG, its applications can be extended to include other biomedical signals, indicating its potential for broader biometric use. [ABSTRACT FROM AUTHOR]

This study addresses the integration of machine learning (ML) with supervisory control and data acquisition (SCADA) systems to enhance predictive maintenance and operational efficiency in oil well monitoring. We investigated the applicability of advanced ML models, including Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), and Momentum LSTM (MLSTM), on a dataset of 21,644 operational records. These models were trained to predict a critical operational parameter, FlowRate, which is essential for operational integrity and efficiency. Our results demonstrate substantial improvements in predictive accuracy: the LSTM model achieved an R2 score of 0.9720, the BiLSTM model reached 0.9725, and the MLSTM model topped at 0.9726, all with exceptionally low Mean Absolute Errors (MAEs) around 0.0090 for LSTM and 0.0089 for BiLSTM and MLSTM. These high R2 values indicate that our models can explain over 97% of the variance in the dataset, reflecting significant predictive accuracy. Such performance underscores the potential of integrating ML with SCADA systems for real-time applications in the oil and gas industry. This study quantifies ML’s integration benefits and sets the stage for further advancements in autonomous well-monitoring systems.