Recurrent Neural Network Method Research Articles

Moving Multi-Object Detection and Tracking Using MRNN and PS-KM Models

On grounds of the advent of real-time applications, like autonomous driving, visual surveillance, and sports analysis, there is an augmenting focus of attention towards Multiple-Object Tracking (MOT). The tracking-by-detection paradigm, a commonly utilized approach, connects the existing recognition hypotheses to the formerly assessed object trajectories by comparing the similarities of the appearance or the motion between them. For an efficient detection and tracking of the numerous objects in a complex environment, a Pearson Similarity-centred Kuhn-Munkres (PS-KM) algorithm was proposed in the present study. In this light, the input videos were, initially, gathered from the MOT dataset and converted into frames. The background subtraction occurred which filtered the inappropriate data concerning the frames after the frame conversion stage. Then, the extraction of features from the frames was executed. Afterwards, the higher dimensional features were transformed into lower-dimensional features, and feature reduction process was performed with the aid of Information Gain-centred Singular Value Decomposition (IG-SVD). Next, using the Modified Recurrent Neural Network (MRNN) method, classification was executed which identified the categories of the objects additionally. The PS-KM algorithm identified that the recognized objects were tracked. Finally, the experimental outcomes exhibited that numerous targets were precisely tracked by the proposed system with 97% accuracy with a low false positive rate (FPR) of 2.3%. It was also proved that the present techniques viz. RNN, CNN, and KNN, were effective with regard to the existing models.

Spatio-temporal prediction of temperature in fluidized bed biomass gasifier using dynamic recurrent neural network method

• Long-short term memory model is proposed to predict temperatures in fluidized bed gasifier. • The proposed model is validated with pilot-scale plant data resolved in space and time. • Multi-step ahead temperature progressions of freeboard, fluidized bed and outlet gas, are predicted. • Gated recurrent unit and long short-term memory models are compared. In this study, a long short-term memory (LSTM) based dynamic recurrent neural network model is proposed for multi-step ahead temperature predictions in a pilot-scale fluidized bed biomass gasifier (FBG). The LSTM model predicts not only the temporal but also the spatial distribution of temperature by considering the temperature of each region of the FBG (fluidized bed, freeboard and outlet gas) as a separate target parameter. The proposed model is validated by comparing simulation data with experimental observations acquired during operation of the FBG. The validation results reveal that the proposed LSTM model is capable of accurately (MAE<6) predicting 1-min-ahead temperature of all the FBG regions. The LSTM model is further challenged for temperature predictions at farther future points (3 min and 5 min ahead) to test the prediction limits of the LSTM model. For 5 min ahead predictions, the proposed LSTM-based prediction model is also compared with other state-of-the-art dynamic neural network methods that include the standard recurrent neural network (S-RNN) and its advanced variant, the gated recurrent unit (GRU). The comparative findings for far future predictions show that LSTM has the highest accuracy, and also exhibit that GRU does not have universally faster convergence than LSTM.

A particle swarm optimisation based recurrent neural network method for predicting the trend of Chinese engineering education reform

A particle swarm optimisation based recurrent neural network method for predicting the trend of Chinese engineering education reform

A big-data-based recurrent neural network method for forest energy estimation

In this paper, the recurrent neural network algorithm of big data is used to conduct in-depth research and analysis on the estimation of forest energy, and the corresponding forestry construction and biomass energy industry development are proposed. Firstly, by introducing a non-negative variable, the bounded constrained underdetermined linear system is transformed into a time-varying system consisting of linear and non-linear equations. Meanwhile, the actual situation of industrial raw material supply of forest biomass is explored by considering the technical level of collection, transportation, and storage. Then, three different alternative scale scenarios are designed to explore the development dynamics of the forest biomass industry in Guizhou under different levels of development and to propose optimization schemes and policy recommendations accordingly. Finally, the simulation results show that in terms of energy efficiency, economic benefits, and environmental benefits, the proposed mehtod can achieve carbon dioxide emission reduction benefits of about 46.17% and increase the economic benefits of carbon dioxide trading by 35.69%.

Deteksi Serangan pada Jaringan Kompleks IoT menggunakan Recurrent Neural Network

The complex network in the Internet of Things is challenging to maintain network security. With network complexity including data, protocols, sizes, communications, standards, and more, it becomes difficult to implement an intrusion detection system (IDS). One way to improve IDS on complex IoT networks is by using deep learning to detect attacks that occur on complex IoT networks. Recurrent neural network (RNN) is a deep learning method that enhances the detection of complex IoT networks because it takes into account the current input as well as what has been learned from previously received inputs. When making decisions about RNNs, consider current information as well as what has been learned from previous input. Therefore, this study proposes the RNN method to improve the performance of attack detection systems on complex IoT networks. The results of this experiment show satisfactory results by increasing the performance of the accuracy detection system in complex IoT networks which reaches 87%.

Extracting Tea Plantations from Multitemporal Sentinel-2 Images Based on Deep Learning Networks

Tea is a special economic crop that is widely distributed in tropical and subtropical areas. Timely and accurate access to the distribution of tea plantation areas is crucial for effective tea plantation supervision and sustainable agricultural development. Traditional methods for tea plantation extraction are highly dependent on feature engineering, which requires expensive human and material resources, and it is sometimes even difficult to achieve the expected results in terms of accuracy and robustness. To alleviate such problems, we took Xinchang County as the study area and proposed a method to extract tea plantations based on deep learning networks. Convolutional neural network (CNN) and recurrent neural network (RNN) modules were combined to build an R-CNN model that can automatically obtain both spatial and temporal information from multitemporal Sentinel-2 remote sensing images of tea plantations, and then the spatial distribution of tea plantations was predicted. To confirm the effectiveness of our method, support vector machine (SVM), random forest (RF), CNN, and RNN methods were used for comparative experiments. The results show that the R-CNN method has great potential in the tea plantation extraction task, with an F1 score and IoU of 0.885 and 0.793 on the test dataset, respectively. The overall classification accuracy and kappa coefficient for the whole region are 0.953 and 0.904, respectively, indicating that this method possesses higher extraction accuracy than the other four methods. In addition, we found that the distribution index of tea plantations in mountainous areas with gentle slopes is the highest in Xinchang County. This study can provide a reference basis for the fine mapping of tea plantation distributions.

Performance Comparison of Deep Learning Algorithm for Speech Emotion Recognition

One of the problems in Speech emotion recognition is related to time series data, while the feedforward process in neural networks is unidirectional where the results from one layer are directly channeled to the next layer. This kind of feedforward process cannot store past data. Thus, if Deep Neural Network (DNN) is used for Speech emotion recognition, some problems arise, such as the speech rate of the speaker. DNN cannot analyze the existing acoustic patterns and so cannot map different levels of speech rate. Another method that can take input at once while retaining relevant data in the previous process is the Recurrent Neural Network (RNN). This paper presents the characteristics of the RNN method consisting of LSTM and GRU techniques for Speech emotion recognition using the Berlin EMODB dataset. The dataset is divided into 80% for training and 20% for testing. The feature extraction methods used are Zero crossing Rate (ZCR), Mel Frequency Cepstral Coefficients (MFCC), Root Mean Square Energy (RMSE), Mel Spectrogram, and Chroma. This study compares the CNN, LSTM, and GRU algorithms. The classification results show that the CNN algorithm gets better results, namely 79.13%. Meanwhile, LSTM and GRU only got an accuracy of 55.76% and 55.14%, respectively

Fostering deep learning approaches to evaluate the impact of urbanization on vegetation and future prospects

Vegetation is an essential component of our global ecosystem and an important indicator of the dynamics and productivity of land cover. Vegetation forecasting research has been accelerated using several deep learning (DL) algorithms through remote sensing (RS) data. In this context, we used artificial intelligence (AI) and the long-short-term memory recurrent neural network (LSTM-RNN) method to explore and forecast future urban–rural vegetation disparities (ΔEVI, where EVI is the enhanced vegetation index) in Pakistan’s six megacities using MODIS EVI data. The forecast results revealed that ΔEVI is decreasing in all cities. The Root Mean Square Error (RMSE) and the Mean Absolute Percentage Error (MAPE) were used to evaluate LSTM-RNN. RSME values were recorded as 0.03, 0.07, 0.02, 0.03, 0.05, and 0.06 for Faisalabad, Gujranwala, Rawalpindi, Lahore, Multan, and Sialkot, respectively. MAPE was estimated as 0.12, 0.55, 0.24, 0.18, 0.28, and 0.47 for Faisalabad, Gujranwala, Rawalpindi, Lahore, Multan, and Sialkot, respectively. This situation indicates that LSTM-RNN can be used as a new reliable AI technique for forecasting. The results suggested that the average of forecasted ΔEVI for the next 10 years is −0.23, −0.21, −0.09, −0.13, −0.22, and −0.11 for Faisalabad, Gujranwala, Rawalpindi, Lahore, Multan, and Sialkot, respectively. The findings of this study will help evaluate the impact of urbanization on EVI by leveraging DL techniques along with implementing an urbanization policy for urban development and environmental protection for long-term urban sustainability.

Multi-dimensional shared representation learning with graph fusion network for Session-based Recommendation

The Session-based Recommendation (SBR) system aims to forecast anonymous users’ short-term decisions. Many prior research have demonstrated that using Graph Neural Networks (GNN) and Recurrent Neural Networks (RNN) to solve SBR tasks can lead to excellent results. However, the existing SBR models only use single feature extraction method to represent an item in the recommendation process. These models either use the GNN methods which ignore the sequence location features in the training process or use RNN methods which cannot obtain the weight features of every item in the sequence. All of them fail to extract features of different dimensions at the same time, which results in low-quality item representations. To tackle the aforementioned issue, this research introduces a novel method called Multi-dimensional Shared Representation Learning (MSR). (i) To get multi-dimensional features, we use the multi-dimensional feature extraction in a double-flow way based on transformer layers and graph attention network (GAT) layers. (ii) Through the MSR module, we train the joint representation of items in the session by using multiple GNNs, then merge the multi-dimensional features by using an attention mechanism. A variety of experiments have been carried out and the simulation results demonstrate the designed strategy largely surpasses the state-of-the-art proposal in recommendation accuracy.

Comparison of Adaptive Holt-Winters Exponential Smoothing and Recurrent Neural Network Model for Forecasting Rainfall in Malang City

Rainfall forecast is necessary for many aspects of regional management. Prediction of rainfall is useful for reducing negative impacts caused by the intensity of rainfall, such as landslides, floods, and storms. Hence, a rainfall forecast with good accuracy is needed. Many rainfall forecasting models have been developed, including the adaptive Holt-Winters exponential smoothing method and the Recurrent Neural Network (RNN) method. The research aimed to compare the result of forecasting between the Holt-Winters adaptive exponential smoothing method and the Recurrent Neural Network (RNN) method. The data were monthly rainfall data in Malang City from January 1983 to December 2019 obtained from a website. Then, the data were divided into training data and testing data. Training data consisted of rainfall data in Malang City from January 1983 to December 2017. Meanwhile, the testing data were rainfall data in Malang City from January 2018 to December 2019. The comparison result was assessed based on the values of Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE). The result reveals that the RNN method has better RMSE and MAPE values, namely RMSE values of 0,377 and MAPE values of 1,596, than the Holt-Winter Adaptive Exponential Smoothing method with RMSE values of 0,500 and MAPE values of 0,620. It can be concluded that the non-linear model has better forecasting than the linear model. Therefore, the RNN model can be used in modeling and forecasting trend and seasonal time series.

Analysis Sentiment On Airline Customer Saisfaction Using Reccurent Neural Network

When talking about customer satisfaction, Twitter as a large and great media could be used to get sentiment or opinion on a product and service of a business. The sentiment will be in a form of tweet that was posted on Twitter that referred to hot debated issues subjectively. The tweet data then will be processed using machine learning to analyze the sentiment of a certain topic. This study aimed to analyze the sentiment of Indonesian public on one of the Indonesian airlines using Deep Learning, Recurrent Neural Network (RNN) method based on the training for Long Short-Term Memory (LSTM), validation and prediction. The tweet will be selected in the span of three years (2017-2020) through the triangulation sentence sentiment process. The LSTM model gives a result of 98.5% accuracy and 92.2% validation accuracy in the data training. Whereas, the LSTM model’s data testing gives a result of 56.5% negative sentiment higher than the positive and neutral sentiment. It could be assumed that the factors which affect the negative sentiment could be used as an input to improve any business process

Development of a Data-Driven Method for Online Battery Remaining-Useful-Life Prediction

Remaining-useful-life (RUL) prediction of Li-ion batteries is used to provide an early indication of the expected lifetime of the battery, thereby reducing the risk of failure and increasing safety. In this paper, a detailed method is presented to make long-term predictions for the RUL based on a combination of gated recurrent unit neural network (GRU NN) and soft-sensing method. Firstly, an indirect health indicator (HI) was extracted from the charging processes using a soft-sensing method that can accurately describe power degradation instead of capacity. Then, a GRU NN with a sliding window was applied to learn the long-term performance development. The method also uses a dropout and early stopping method to prevent overfitting. To build the models and validate the effectiveness of the proposed method, a real-world NASA battery data set with various battery measurements was used. The results show that the method can produce a long-term and accurate RUL prediction at each position of the degradation progression based on several historical battery data sets.

IoT-Based Intelligent Monitoring System Applying RNN

In this paper, we propose an intelligent monitoring framework based on the Internet of Things (IoT) by applying a Recurrent Neural Network (RNN) for the predictive maintenance of a biobanking system. RNN, which is one of the deep learning models, is used for time series data. It is called a sequence model because it processes inputs and outputs in sequence units. The proposed framework measures the internal temperature of the cryogenic freezer and the temperature of each component simultaneously, monitors the internal temperatures of internal and middle layers in real time, sends the sensing temperature data to the server, and performs predictive learning. Thus, it is possible to support the intelligent predictive maintenance of the biobank by performing a time series data analysis of the temperature sensor using RNN. Among RNN methods, a simple RNN has a longer-term dependency problem; therefore, Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU), which have higher learning performance, are selected. To support the intelligent predictive maintenance of the biobank, both the LSTM and GRU models were constructed, and comparative experiments were performed. The proposed system can ensure the safety of bio-resources by performing predictive maintenance using RNN and provide an accurate status of the biobank in real-time. In addition, before an abnormal situation occurs, it is possible to respond immediately to emergencies that may damage biological resources.

Cardiac disease classification from ecg signals using hybrid recurrent neural network method

Medical practitioners and cardiologists frequently employ electrocardiogram (ECG) analysis to assess cardiac health. An automated ECG categorization systems have difficulty detecting and analysing the signals of numerous waves of populations, particularly electrocardiogram (ECG) data. A variety of learning strategies have been researched for the interpretation of ECG categorization. One drawback of machine learning is the use of heuristic features with shallow feature learning architectures. In order to overcome this difficulty, a deep learning approach is used to automatically learn features as opposed to using conventional handmade features. In this work, the use of Long Short-Term Memory (LSTM) and Gated Recurrent Neural Network (GRNN) techniques is proposed to develop an accurate ECG categorization and monitoring system. The purpose of the gated recurrent neural network to processing the sequential data and check the error presented or not in the ECG signals. The long short-term memory is the type of the neural network approaches to identify the complex sequence prediction problems. The GRNN and LSTM models then receive the learnt features that were first captured by the CNN model. There are no custom features required for the ECG categorization in the model. The results section lists various cutting-edge models that the CNN-LSTM and CNN-GRNN models outperformed. The CNN-LSTM and CNN-GRNN designs outperform conventional ECG classification architectures with comparable hyper-parameters, according to the comparison results.

The Organization Entity Extraction Telkom University Affiliated using Recurrent Neural Network (RNN)

In the news portal text, there is a lot of important information such as the name of the person, the name of the organization, or the name of the place. To obtain information in text documents manually, humans must read the contents of the entire news text. To overcome this issue, information extraction, commonly called Named Entity Recognition (NER) was used. The extraction of information expressly for the NER field is used to make it easier to process word or sentence data. It helps search engines and also helps to classify places, times, and organizations. There is a limited number of NER in Indonesian texts using only the Recurrent Neural Network (RNN) method. Similar previous studies only employed other versions of RNN such as LSTM (Long Short Term Memory), BiLSTM (Bidirectional Long Short Term Memory), and CNN (Convolutional Neural Network). NER is one of the answers to the problems that exist in a large number of news portal texts to obtain information effectively and efficiently. The results of this study indicate that the NER system using the RNN method in Indonesian news texts has an F1 -Score of 80%

Visual Field Prediction: Evaluating the Clinical Relevance of Deep Learning Models

Two novel deep learning methods using a convolutional neural network (CNN) and a recurrent neural network (RNN) have recently been developed to forecast future visual fields (VFs). Although the original evaluations of these models focused on overall accuracy, it was not assessed whether they can accurately identify patients with progressive glaucomatous vision loss to aid clinicians in preventing further decline. We evaluated these 2 prediction models for potential biases in overestimating or underestimating VF changes over time. Retrospective observational cohort study. All available and reliable Swedish Interactive Thresholding Algorithm Standard 24-2 VFs from Massachusetts Eye and Ear Glaucoma Service collected between 1999 and 2020 were extracted. Because of the methods' respective needs, the CNN data set included 54 373 samples from 7472 patients, and the RNN data set included 24 430 samples from 1809 patients. The CNN and RNN methods were reimplemented. A fivefold cross-validation procedure was performed on each model, and pointwise mean absolute error (PMAE) was used to measure prediction accuracy. Test data were stratified into categories based on the severity of VF progression to investigate the models' performances on predicting worsening cases. The models were additionally compared with a no-change model that uses the baseline VF (for the CNN) and the last-observed VF (for the RNN) for its prediction. PMAE in predictions. The overall PMAE 95% confidence intervals were 2.21 to 2.24 decibels (dB) for the CNN and 2.56 to 2.61 dB for the RNN, which were close to the original studies' reported values. However, both models exhibited large errors in identifying patients with worsening VFs and often failed to outperform the no-change model. Pointwise mean absolute error values were higher in patients with greater changes in mean sensitivity (for the CNN) and mean total deviation (for the RNN) between baseline and follow-up VFs. Although our evaluation confirms the low overall PMAEs reported in the original studies, our findings also reveal that both models severely underpredict worsening of VF loss. Because the accurate detection and projection of glaucomatous VF decline is crucial in ophthalmic clinical practice, we recommend that this consideration is explicitly taken into account when developing and evaluating future deep learning models.

Multi-source rainfall merging and reservoir inflow forecasting by ensemble technique and artificial intelligence

Study regionThe Shihmen reservoir, the second largest reservoir in northern Taiwan, is flooded frequently, and featured by short period of flood peak due to uneven distribution of rainfall and mountainous topography. Study focusWe proposed a novel streamflow-oriented ensemble recurrent neural networks (ERNN) method to merge three precipitation products, namely gauge measurements, radar and satellite rainfall products. We analyzed (1) the effect of artificial intelligence method for bias correction of precipitation products with reference to streamflow, (2) the comparison of two widely used arithmetic average (AA) and Bayesian model averaging (BMA) methods with ERNN, (3) the performance of merged rainfalls and the original three precipitation products in inflow forecasting during 13 typhoon events. New hydrological insightsWe found that all precipitation products are biased and can be appropriately adjusted with an improvement in inflow forecasting of about 2.5 %, 21.8 % and 60.7 % for gauge, radar and satellite rainfall products in terms of RMSE. After rainfall merging, RMSE for radar and satellite are further improved by 14% and 36% at least. ERNN merging method indicates that the optimal merging weights for gauge, radar and satellite are within the range of 0.52–0.56, 0.35–0.39 and 0.08–0.09, respectively according to 95 % confidence interval. The merged rainfall product skillfully predicts the inflow with a lead time of five hours, and ERNN merging method is superior to traditional AA and BMA methods, with NS up to 0.85 and CRPS reduced by near 50 % compared to BMA. Additionally, ERNN is found to capture the peak times and peak flows with the smallest error. Overall, this study provides a potential approach to merge multiple rainfall products and to obtain the effective rainfall by fitting the hydrological responses over mountainous watersheds, where observational biases may frequently occur in gauge, radar and satellite measurements.

Evaluation of artificial neural network methods to forecast short-term solar power generation: a case study in Eastern Mediterranean Region

Solar power forecasting is substantial for the utilization, planning, and designing of solar power plants. Global solar irradiation (GSI) and meteorological variables have a crucial role in solar power generation. The ever-changing meteorological variables and imprecise measurement of GSI raise difficulties for forecasting photovoltaic (PV) output power. In this context, a major motivation appears for the accurate forecast of GSI to perform effective forecasting of the short-term output power of a PV plant. The presented study comprises of four artificial neural network (ANN) methods; recurrent neural network (RNN) method, feedforward backpropagation neural network (FFBPNN) method, support vector regression (SVR) method, and long short-term memory (LSTM) for daily total GSI prediction of Tarsus by using meteorological data. Moreover, this study proposes a model that utilizes the predicted daily GSI for output power forecasting of a grid-connected PV plant. The obtained results are compared with the output power generation data of a 350 kW solar power plant. The results are evaluated with the performance indices as mean absolute percentage error (MAPE), normalized root mean squared error (NRMSE), weighted mean absolute error (WMAE), and normalized mean absolute error (NMAE). FFBPNN method is chosen with the best results of MAPE 7.066%, NMAE 3.629%, NRMSE 4.673%, and WMAE 5.256%.

Sentiment Analysis of Student Satisfaction on Telkom University Language Center (LaC) Services on Instagram Using the RNN Method

Social media has become a medium for communication between individuals and aspects of the business, including decision-making processes, brand promotion, brand marketing, and personal branding. One of them is Instagram. Using the comments feature on Instagram, users can communicate and give opinions on an upload on an Instagram account. Sentiment analysis can be done to analyze comments on the LaC (language center) Instagram account to measure student satisfaction sentiment towards Telkom university's LaC (language center) services. This study aims to analyze the sentiment or opinion of student satisfaction with the Telkom University Language Center (LaC) service on Instagram. The author also performs a classification based on positive sentiment, negative, and neutral categories using the Recurrent Neural Network (RNN) method and the Confusion Matrix measurement. From the test results on the model built to get an accuracy value of 79%.

A Gaussian High-Dimensional Random Matrix-Based Method for Detecting Abnormal Student Behaviour in Chinese Language Classrooms

In this paper, a Gaussian high-dimensional random matrix approach is used to conduct an in-depth study and analysis of the detection of abnormal student behaviour in Chinese language classrooms, and a corresponding model is designed for practical application. The Gaussian high-dimensional random matrix technique and recurrent neural network technique are applied to build a basic technical framework for intelligent proctoring. An innovative method of human posture estimation is used to complete the proctoring task, and the structure of the Gaussian high-dimensional random matrix method is simplified and the model is compressed to achieve real-time and parallelism of the method. A PolSAR sparse representation classification (OGRM_SRC) algorithm based on orthogonal Gaussian random matrices (OGRM) is proposed to construct an observation dictionary based on polarisation features and OGRM to select typical ground target samples in polarized feature images. The system not only functions as a psychometric assessment of students but also is capable of detecting and analysing abnormal student behaviour based on the improved forest algorithm. The residuals of the pixel to be classified are calculated relative to each atom in the observation dictionary, and the minimum reconstruction residual is used as the classification criterion. The OGRM_SRC algorithm is applied to classify the pixel, and the final classification results are obtained and evaluated for accuracy. We propose a fine-grained action recognition optimisation method for recurrent neural networks, fusing temporal attention information and spatial attention information to improve the action recognition method and improve the model’s ability to recognise fine-grained abnormal behaviours such as looking left, looking right, and copying with head down. Based on this, we conducted some comparative experiments to validate the effectiveness of our work and verified that the system achieved a recognition efficiency of 76.8 FPS and a recognition accuracy of 98.5% in a graphical computing processor.