Bi-LSTM Model Research Articles

Sentiment Analysis on IMDB Review Dataset

A computational method known as sentiment analysis is employed to ascertain the emotional undertone or attitude of a text document, such as a review, tweet, or news story. Using machine learning models, deep neural network models, and natural language processing, the method entails examining the text to determine whether it expresses positive or negative sentiment. In this study, models like Naive Bayes, Logistic Regression, LSTM, LSVM, Decision tree, and BiLSTM are utilized to conduct a sentiment analysis (SA) study on the IMDB dataset. The goal of the investigation is to evaluate how well these models perform in retrospect on movie reviews, categorizing them as positive or negative. The study investigates the effects of data pre-processing methods and hyperparameter tuning on the models’ accuracy. The final results demonstrate that the BiLSTM model outperforms the other models in terms of recall, precision, and accuracy, followed by the LSTM, Logistic Regression, LSVM, Decision Tree, and Naive Bayes models. The research emphasizes the potential of deep learning models—in particular, BiLSTM in sentiment analysis tasks, as well as the significance of hyper-parameter tuning and pre-processing methods in achieving high accuracy.

Optimization of the visual communication path of data news in the context of the Internet

Abstract Visual news belongs to a kind of data news, which uses emerging technology to bring users a new reading experience and improve the news dissemination effect. In this paper, we utilize the self-attention mechanism and BiLSTM model to extract the content of data news documents and select and classify the audio features of data news by the AdaBoost algorithm. ResNet50 is used as the backbone network, combined with multiple residual unit modules for news multi-scale image feature extraction, and the text, audio and image features of the data news are fused by constructing a multimodal feature adaptive fusion model to further optimize the visual display of the data news. This paper verifies the model’s effectiveness for the data news visualization technique through multiple perspectives, including training loss curve, performance comparison, and ablation experiment. The results show that the loss value of the model based on the Self-Attention-BiLSTM model for news text feature extraction is 0.211 after 150 iterations. The AdaBoost-based news audio feature classification method has an average classification accuracy of 97.56% for the positive diagonal when the noise is raised from 0dB to 30 dB. By adding the residual unit module to the ResNet50 backbone network, the accuracy of the multi-scale image feature extraction model has been improved by 2.77% compared to the single backbone network. The information age requires the full integration of technology and news and combining the corresponding Internet platform for the visual display of data news in order to promote the expansion of the visual communication path of data news.

Predicting Traffic Flow Parameters for Sustainable Highway Management: An Attention-Based EMD–BiLSTM Approach

The long-term prediction of highway traffic parameters is frequently undermined by cumulative errors from various influencing factors and unforeseen events, resulting in diminished predictive accuracy and applicability. In the pursuit of sustainable highway development and eco-friendly transportation strategies, forecasting these traffic flow parameters has emerged as an urgent concern. To mitigate issues associated with cumulative error and unexpected events in long-term forecasts, this study leverages the empirical mode decomposition (EMD) method to deconstruct time series data. This aims to minimize disturbances from data fluctuations, thereby enhancing data quality. We also incorporate the BiLSTM model, ensuring bidirectional learning from extended time series data for a thorough extraction of relevant insights. In a pioneering effort, this research integrates the attention mechanism with the EMD–BiLSTM model. This synergy deeply excavates the spatiotemporal characteristics of traffic volume data, allocating appropriate weights to significant information, which markedly boosts predictive precision and speed. Through comparisons with ARIMA, LSTM, and BiLSTM models, we demonstrate the distinct advantage of our approach in predicting traffic volume and speed. In summary, our study introduces a groundbreaking technique for the meticulous forecasting of highway traffic volume. This serves as a robust decision-making instrument for both sustainable highway development and transportation management, paving the way for more sustainable, efficient, and environmentally conscious highway transit.

Deep Learning Based Automatic Answer Scoring Through Bi-Directional LSTM

Due to population expansion and the increasing importance of education, it is becoming increasingly difficult for assessors to evaluate the correctness and relevance of the responses provided by students. The LSTM model was initially used to build the answer-scoring system. The Bi-LSTM model has been designed with callbacks to acquire the student answer scoring system due to the LSTM's limitations for optimal scoring. The proposed system has been implemented using the ASAP Short Answer Scoring dataset. The results show that the system developed using Bi-LSTM displays better performance than LSTM.

A Machine Learning Scheme for Speed Prediction in Intelligent Transportation Systems Using a Bi-LSTM Based Model

Congestion has become a big issue in major cities throughout the world. Numerous transportation activities have been affected, and travel times have increased as a result of many travelers spending lengthy hours on the road. To overcome these challenges, the Intelligent Transportation System (ITS), which provides efficient traffic service and management, has sparked widespread attention. Collection and analysis of traffic data has been made possible by the algorithms implemented by the ITS. Huge volumes of data are produced by the vast wide range of sensors used in the Internet of Things (IoT), enabling the collection of a variety of traffic information. Development of short-term traffic speed prediction, has been made possible using deep learning models such as Long Short Term Memory (LSTM) and Bidirectional LSTM. Numerous variables, including the weather, the state of the roads, and traffic congestion, can have long-term dependencies and influence traffic speed. The bidirectional architecture of Bi-LSTMs enables them to handle long-term dependencies in sequences and efficiently capture both past and future context in a sequence, which is crucial for producing accurate forecasts of traffic speed. In this paper, the upstream and downstream flow of traffic speed on various pathways has been investigated using a traffic path planning algorithm based on Bi-LSTM models. The algorithm considers the factors affecting the flow of traffic at different seasons and time of the day and tries to predict the average speed associated to that path several timeslots ahead. The experimental results demonstrated that the Bi-LSTM model has the benefit of predicting speed for various timeslots while retaining a high level of accuracy.

Intelligent big data visual analytics based on deep learning

Abstract In this paper, we first constructed a deep learning model, optimized the LSTM model to get the BiLSTM model based on the long and short-term memory network, and used the generative adversarial network to calculate the probability distribution of data. Then, the advantages of deep learning in intelligent big data visualization and analysis are explored from the dimensions of data preprocessing, dimension anchor layout, coordinate expansion and data analysis. Finally, the efficiency of the deep learning model is compared with that of other algorithms using indicators such as accuracy and recall, and the feasibility of this paper’s method is verified by empirical analysis using intelligent transportation data as an example. The results show that the model in this paper achieves an accuracy rate of 95.5%, the loss rate is stable at 0.2% to 0.4%, and the average running time is maintained at 20ms, which are all better than other models. The predicted and real values of traffic data for the Deep-STCL model using deep learning basically match, indicating that the deep learning model has obvious advantages in data visualization and analysis.

Text Sentiment Classification Method Based on Bilstm

In the booming age of digital communication and natural language processing(NLP), understanding user sentiments expressed in texts has become increasingly vital. This research addresses the challenge of classifying text sentiments into three distinct categories: positive, neutral, and negative. Recognizing the nuances of human emotion in textual content is paramount for enhancing user experiences and tailoring personalized digital interactions. This study trains a BiLSTM model, supplemented with optimization strategies to adaptively learn from a diverse dataset. Additionally, a large number of baseline experiments were conducted to compare its efficacy with traditional algorithms such as Logistic Regression. The BiLSTM model demonstrated a good accuracy of 78.21% on the test set. Conclusively, while the proposed model shows significant potential, further refinements could be made, emphasizing data augmentation and model regularization to achieve optimal performance.

A Landslide Displacement Prediction Model Based on the ICEEMDAN Method and the TCN–BiLSTM Combined Neural Network

Influenced by autochthonous geological conditions and external environmental changes, the evolution of landslides is mostly nonlinear. This article proposes a combined neural network prediction model that combines a temporal convolutional neural network (TCN) and a bidirectional long short-term memory neural network (BiLSTM) to address the shortcomings of traditional recurrent neural networks in predicting displacement-fluctuation-type landslides. Based on the idea of time series decomposition, the improved complete ensemble empirical mode decomposition with an adaptive noise method (ICEEMDAN) was used to decompose displacement time series data into trend and fluctuation terms. Trend displacement is mainly influenced by the internal geological conditions of a landslide, and polynomial fitting is used to determine the future trend displacement; The displacement of the fluctuation term is mainly influenced by the external environment of landslides. This article selects three types of landslide-influencing factors: rainfall, groundwater level elevation, and the historical displacement of landslides. It uses a combination of gray correlation (GRG) and mutual information (MIC) correlation modules for feature screening. Then, TCN is used to extract landslide characteristic factors, and BiLSTM captures the relationship between features and displacement to achieve the prediction of wave term displacement. Finally, the trend term and fluctuation term displacement prediction values are reconstructed to obtain the total displacement prediction value. The results indicate that the ICEEMDAN–TCN–BiLSTM model proposed in this article can accurately predict landslide displacement and has high engineering application value, which is helpful for planning and constructing landslide disaster prevention projects.

Reactive Power Output Modeling of Synchronous Condenser in UHVDC Converter Station Based on Interlaced Superposition CNN-BiLSTM

To guarantee stable power system operation, a synchronous condenser (SC) is configured in an ultra-high voltage direct current (UHVDC) converter station to provide dynamic reactive power support to the power system. The research on the reactive power output modelling of a SC in an UHVDC converter station has important theoretical significance and practical value for the reactive power control of a SC in an UHVDC converter station. Focusing on the reactive power regulation system of the SC with strong coupling, multivariable, and nonlinear features, it is difficult for the universal analytic method to build the SC reactive power output model. A novel reactive power output model of the SC in the UHVDC converter station based on interlaced superposition convolutional neural network-bidirectional long short-term memory (CNN-BiLSTM) is proposed, in which a novel interlaced superposition CNN composed of convolution units with two different structures is built to increase the depth of the network and avoid over-fitting. Particularly the branch channels of convolution units with two different structures are connected by a convolution layer and a skip connection respectively. The interlaced superposition CNN-BiLSTM model is that the combination of the interlaced superposition CNN model and BiLSTM model for improving the model accuracy and generic capability. The Bayesian optimization method is used to optimize its hyperparameters. The application of interlaced Superposition CNN-BiLSTM in SC reactive power output modelling is a new technology. The excitation voltage and excitation current of the SC are used as inputs for the training and testing sampled data, and the reactive power of the SC is used as outputs for the training and testing sampled data. Therefore, the reactive power output model of a SC in an UHVDC converter station based on interlaced superposition CNN-BiLSTM is obtained, and it attains a low root mean square error (RMSE = 0.126750) and a high determination coefficient (R2 = 0.999999).

Predictive Power of XGBoost_BiLSTM Model: A Machine-Learning Approach for Accurate Sleep Apnea Detection Using Electronic Health Data

Sleep apnea is a common disorder that can cause pauses in breathing and can last from a few seconds to several minutes, as well as shallow breathing or complete cessation of breathing. Obstructive sleep apnea is strongly associated with the risk of developing several heart diseases, including coronary heart disease, heart attack, heart failure, and stroke. In addition, obstructive sleep apnea increases the risk of developing irregular heartbeats (arrhythmias), which can lead to low blood pressure. To prevent these conditions, this study presents a novel machine-learning (ML) model for predicting sleep apnea based on electronic health data that provides accurate predictions and helps in identifying the risk factors that contribute to the development of sleep apnea. The dataset used in the study includes 75 features and 10,765 samples from the Swedish National Study on Aging and Care (SNAC). The proposed model is based on two modules: the XGBoost module assesses the most important features from feature space, while the Bidirectional Long Short-Term Memory Networks (BiLSTM) module classifies the probability of sleep apnea. Using a cross-validation scheme, the proposed XGBoost_BiLSTM algorithm achieves an accuracy of 97% while using only the six most significant features from the dataset. The model’s performance is also compared with conventional long-short-term memory networks (LSTM) and other state-of-the-art ML models. The results of the study suggest that the proposed model improved the diagnosis and treatment of sleep apnea by identifying the risk factors.

Investor Sentiment Mining Based on Bi-LSTM Model and its Impact on Stock Price Bubbles

Abstract We built a Bi-Directional long-term and short-term memory (Bi-LSTM) model to identify and classify the Chinese posting text of stocks on the Eastmoney website in China and constructed the daily index of Chinese investors’ sentiment. Furthermore, based on the GSADF method, we examine the stock price bubbles and study the impact of investor sentiment and stock price bubbles. We found investor sentiment has a positive effect on the existence of stock bubbles, as well as their intensity. This effect is more significant in small-scale, high-equity concentration, and non-state-owned enterprises. Investor sentiment has an impact on stock price bubbles through volatility, and stock price bubbles are often accompanied by higher premium risk. The conclusion is helpful to understand the mechanism of investor sentiment on stock bubbles from a micro perspective, and it also can be a reference in identifying stock bubbles from the viewpoint of investor sentiment.

Neural network-based prediction system for port throughput: A case study of Ningbo-Zhoushan Port

Port throughput is a crucial indicator for assessing the efficiency of port enterprises. Accurately predicting long-term port throughput can help mitigate the risk of investing in major infrastructure during the development process. This study introduces a neural network-based system for predicting port throughput at the Ningbo-Zhoushan Port. By utilizing the grey relational analysis method, we identified 12 highly correlated factors that affect port throughput variation in Ningbo-Zhoushan Port. These factors were then used as input indicators for the BiLSTM model to predict port throughput. Our research indicates that the proposed model achieved a high level of accuracy, with a MAPE value of 2.95%, outperforming commonly used machine learning and neural network models. Additionally, the model's predictions suggest a rapid growth trend in cargo throughput at Ningbo-Zhoushan Port by 2026. Consequently, we provide relevant suggestions to address this trend. Our research can assist port enterprises in making informed decisions, contributing to the field of port logistics.

A dual-optimization wind speed forecasting model based on deep learning and improved dung beetle optimization algorithm

Accurate wind speed forecasting is capable of increasing the stability of wind power system. Notably, there are numerous factors affecting wind speed, thus causing wind speed forecasting to be difficult. To address the above-mentioned challenge, a novel hybrid model integrating genetic algorithm (GA), variational mode decomposition (VMD), improved dung beetle optimization algorithm (IDBO), and Bidirectional long short-term memory network based on attention mechanism (BiLSTM-A) is proposed in this study to achieve satisfactory forecasting performance. In the proposed model, GA is adopted to optimize the VMD to eliminate noise and extract original series attributes. And the IDBO is adopted for hyperparameters selection for the BiLSTM-A. The proposed GA-VMD-IDBO-BiLSTM-A is compared with nine established comparable models, with the aim of verifying its forecasting performance. A series of experiments on four 1-hour real wind series in Stratford are performed to assess the performance of the model. The MAPE of the four datasets forecasting results reached 1.4%, 2.4%, 3.5%, 2.4%. As indicated by the experimental results, GA-VMD can better process the data and improve the forecasting accuracy. IDBO can optimize the parameters of BiLSTM model and improve the forecasting performance. The dual-optimization wind speed forecasting model can obtain high accuracy and strong stability.

Estimation of internal states in a Li-ion battery using BiLSTM with Bayesian hyperparameter optimization

An accurate state estimator is critical to Battery Management Systems (BMSs). In more advanced BMSs, state estimators based on electrochemical-thermal battery models are preferred because they can predict the internal states of the battery. However, these physics-based models are computationally costly and time-demanding, preventing them from being directly used for real-time estimation. As an alternative approach, data-driven models, like Deep Neural Networks (DNNs), can be trained on datasets generated by electrochemical models. Increasing the accuracy of data-driven models can be achieved by applying more complex architectures of DNN. However, such DNNs are computationally time-demanding, which makes them inefficient for online state estimation algorithms. To address this issue, we have proposed a BiLSTM model as a state estimator whose hyperparameters are automatically optimized using a Bayesian Optimization (BO) framework. We demonstrate that by applying Bayesian inference, one can have a highly accurate state estimator while using a less complex DNN architecture which allows for being computationally efficient as well. The high accuracy of the BiLSTM model with the optimized combination of hyperparameters is demonstrated by the low Root Mean Squared Errors (RMSE) levels on the test data set.

Estimation of walnut crop evapotranspiration under different micro-irrigation techniques in arid zones based on deep learning sequence models

To solve the problems of low accuracy of crop evapotranspiration (ETc) estimates in arid areas and to optimise ETc for precision irrigation in agriculture.Taking walnut in Wensu County, Aksu Region, Xinjiang as the research object, based on the multivariate time series data of walnut fertility, deep learning sequence models such as LSTM, GRU and BiLSTM were utilized to estimate the crop evapotranspiration (ETc) of walnut in arid zones under different micro-irrigation techniques. The results showed that the coefficient of determination (R2) of the LSTM, GRU and BiLSTM models were up to greater than 0.95 for both microirrigation techniques, and the BiLSTM model was relatively optimal. For the 3 g microirrigation technique, the R2, RMSEV and RPDV of the BiLSTM model were 0.978, 0.294 and 6.965 mm/d respectively. For the Hg microirrigation technique, the R2, RMSEV and RPDV of the BiLSTM model were 0.970, 0.339 and 6.026 mm/d respectively.With different scenarios of missing meteorological environmental variables, the BiLSTM model was used to estimate R2 > 0.98 for walnut crop evapotranspiration under different microirrigation techniques, with a range from 0.958 to 0. 986 under 3 g microirrigation technique, RMSEV ranged from 0.235 to 0.412 and RPDV ranged from 5.028 to 8.494. Under Hg microirrigation technique, R2 ranged from 0.942 to 0.979, RMSEV ranged from 0.288 to 0.477 and RPDV ranged from 4.253 to 6.905. Using the deep learning BiLSTM model, we can efficiently estimate the evapotranspiration of the walnut crop in arid zones with different micro-irrigation technologies. The evapotranspiration can provide a scientific theoretical reference for intelligent irrigation of walnut and agricultural water management.

A Study on the Prediction of COVID-19 Confirmed Cases Using Deep Learning and AdaBoost-Bi-LSTM Model

In this study, AdaBoost-Bi-LSTM ensemble models are developed to predict the number of COVID-19 confirmed cases by effectively learning volatile and unstable data using a nonparametric method. The performance of the developed models in terms of prediction accuracy is compared with those of existing deep learning models such as GRU, LSTM, and Bi-LSTM. The COVID-19 outbreak in 2019 has resulted in a global pandemic with a significant number of deaths worldwide. There have long been ongoing efforts to prevent the spread of infectious diseases, and a number of prediction models have been developed for the number of confirmed cases. However, there are many variables that continuously mutate the virus and therefore affect the number of confirmed cases, which makes it difficult to accurately predict the number of COVID-19 confirmed cases. The goal of this study is to develop a model with a lower error rate and higher predictive accuracy than existing models to more effectively monitor and handle endemic diseases. To this end, this study predicts COVID-19 confirmed cases from April to October 2022 based on the analysis of COVID-19 confirmed cases data from 16 December 2020 to 27 September 2022 using the developed models. As a result, the AdaBoost-Bi-LSTM model shows the best performance, even though the data from the period of high variability in the number of confirmed cases was used for model training. The AdaBoost-Bi-LSTM model achieved improved predictive power and shows an increased performance of 17.41% over the simple GRU/LSTM model and of 15.62% over the Bi-LSTM model.

MODEL BIDIRECTIONAL LSTM UNTUK PEMROSESAN SEKUENSIAL DATA TEKS SPAM

This study examines the LSTM-based model for processing spam in text data. Spam poses several dangers and risks, both for individuals and organizations. Spam can be a nuisance that hampers both individual and organizational productivity. Much spam contains fraudulent or phishing attempts to obtain sensitive information. Spam detection using deep learning involves the utilization of algorithms and deep neural network models to accurately classify messages as either spam or not spam. Typically, spam detection systems use a combination of these methods to improve the accuracy of identifying spam messages. This study applies the Bi-LSTM deep learning model to sequentially process text (sequencing). The performance of the model is determined based on the loss and accuracy. The data used are the Spam SMS and Spam Email datasets. The test results show that the Bi-LSTM model demonstrates better performance on all tested datasets. Bi-LSTM is able to capture textual patterns from both the context and the text itself, as it can combine information from both directions. The test results prove that the Bi-LSTM model is more effective in text comprehension.

BDBRC: A Chinese military entity recognition model combining context contribution and residual dilatation convolutional networks

Military information is gradually overloaded due to the diversity of sources and the exponential growth in quantity, which greatly affects the accuracy of intelligence personnel in extracting and analyzing military information. The modern warfare approach has also evolved from the traditional physical domain to the cognitive domain, and competing for advantages in the cognitive domain has become a key objective of combat. Therefore, constructing domain knowledge graphs and mining the relationships between data play an important role in cognitive domain analysis. In this paper, we propose a convolutional network recognition method based on improved two-layer bi-directional BiLSTM networks named the BERT-α BiLSTMs-RECNN-CRF (BDBRC) model. For the difficulties of military entities generally having long names and low extraction accuracy, as well as the existence of a large number of composite entities that are difficult to recognize, an improved two-layer BiLSTM model is devised first. In view of the fact that the BiLSTMs model always extracts features equally in long-distance text sequences without actually considering the different influences of different sentence contexts, contribution factor a is added to extract the contribution of the above and below to the target entity in different sentences respectively. Then, aiming at the strong problem of the domain of military news texts and the high level of inter-entity ambiguity, we propose a method that utilizes a modified convolutional network (RECNN) for partial feature extraction and jointly with a modified two-layer BiLSTM network for entity recognition. The experiment on the self-constructed dataset shows that the F1 value of the model proposed in this paper reaches 93.18%, and the F1 value, P, and H of our model are all better than the baseline model, which verifies the performance of the model. At the same time, we use public data sets MSRA and CLUB2020, and the experimental results show that the model proposed in this paper also has a good performance in the public data set, verifying the universality of the model. It can provide methodological support for the construction of the military knowledge graph.

Predicting energy consumption of chiller plant using WOA-BiLSTM hybrid prediction model: A case study for a hospital building

It is important to study building energy consumption considering the current state of global climate and its close relationship with building energy consumption. This study proposes a hybrid Whale Optimization Algorithm-Bidirectional Long Short-Term Memory (WOA-BiLSTM) model to predict energy consumption using a hospital building located in Beijing as a case study. First, a simulation is performed using building simulation software and the model is calibrated using energy bill and electricity meter data. Subsequently, the energy consumption of the calibrated model is decomposed to determine the energy consumption of the chillers and pumps. Finally, a BiLSTM model is established and the WOA is employed to enhance the generalization and performance of the BiLSTM model. Results show 1) for hour scale, the Mean Absolute Percentage Error (MAPE) of the WOA-BiLSTM model is reduced by 2.93 % and 0.75 % over the WOA-LSTM and BiLSTM models, respectively. 2) the time-lag effect of energy consumption is discussed and the results show that the optimal window size is approximately 18 h. 3) the optimization results of different iterations and population size of WOA show that performance is optimal when 400 iterations and a population size of 5, and larger datasets require more iterations based on a comparison of two datasets (hour and day scales).

Speech emotion recognition and classification using hybrid deep CNN and BiLSTM model

Speech emotion recognition and classification using hybrid deep CNN and BiLSTM model