Bidirectional Long Short-term Memory Layers Research Articles

Deep Learning Based Lipreading for Video Captioning

Visual speech recognition, often referred to as lipreading, has garnered significant attention in recent years due to its potential applications in various fields such as human-computer interaction, accessibility technology, and biometric security systems. This paper explores the challenges and advancements in the field of lipreading, which involves deciphering speech from visual cues, primarily movements of the lips, tongue, and teeth. Despite being an essential aspect of human communication, lipreading presents inherent difficulties, especially in noisy environments or when contextual information is limited. The McGurk effect, where conflicting audio and visual cues lead to perceptual illusions, highlights the complexity of lipreading. Human lipreading performance varies widely, with hearing-impaired individuals achieving relatively low accuracy rates. Automating lipreading using machine learning techniques has emerged as a promising solution, with potential applications ranging from silent dictation in public spaces to biometric authentication systems. Visual speech recognition methods can be broadly categorized into those that focus on mimicking words and those that model visemes, visually distinguishable phonemes. While word-based approaches are suitable for isolated word recognition, viseme-based techniques are better suited for continuous speech recognition tasks. This study proposes a novel deep learning architecture for lipreading, leveraging Conv3D layers for spatiotemporal feature extraction and bidirectional LSTM layers for sequence modelling. The proposed model demonstrates significant improvements in lipreading accuracy, outperforming traditional methods on benchmark datasets. The practical implications of automated lipreading extend beyond accessibility technology to include biometric identity verification, security surveillance, and enhanced communication aids for individuals with hearing impairments. This paper provides insights into the advancements, challenges, and future directions of visual speech recognition research, paving the way for innovative applications in diverse domains.

Predicting Transcription Factor Binding Sites with Deep Learning.

Prediction of binding sites for transcription factors is important to understand how the latter regulate gene expression and how this regulation can be modulated for therapeutic purposes. A consistent number of references address this issue with different approaches, Machine Learning being one of the most successful. Nevertheless, we note that many such approaches fail to propose a robust and meaningful method to embed the genetic data under analysis. We try to overcome this problem by proposing a bidirectional transformer-based encoder, empowered by bidirectional long-short term memory layers and with a capsule layer responsible for the final prediction. To evaluate the efficiency of the proposed approach, we use benchmark ChIP-seq datasets of five cell lines available in the ENCODE repository (A549, GM12878, Hep-G2, H1-hESC, and Hela). The results show that the proposed method can predict TFBS within the five different cell lines very well; moreover, cross-cell predictions provide satisfactory results as well. Experiments conducted across cell lines are reinforced by the analysis of five additional lines used only to test the model trained using the others. The results confirm that prediction across cell lines remains very high, allowing an extensive cross-transcription factor analysis to be performed from which several indications of interest for molecular biology may be drawn.

Instrument Detection and Descriptive Gesture Segmentation on a Robotic Surgical Maneuvers Dataset

Large datasets play a crucial role in the progression of surgical robotics, facilitating advancements in the fields of surgical task recognition and automation. Moreover, public datasets enable the comparative analysis of various algorithms and methodologies, thereby assessing their effectiveness and performance. The ROSMA (Robotics Surgical Maneuvers) dataset provides 206 trials of common surgical training tasks performed with the da Vinci Research Kit (dVRK). In this work, we extend the ROSMA dataset with two annotated subsets: ROSMAT24, which contains bounding box annotations for instrument detection, and ROSMAG40, which contains high and low-level gesture annotations. We propose an annotation method that provides independent labels for the right-handed tools and the left-handed tools. For instrument identification, we validate our proposal with a YOLOv4 model in two experimental scenarios. We demonstrate the generalization capabilities of the network to detect instruments in unseen scenarios. On the other hand, for gesture segmentation, we propose two label categories: high-level annotations that describe gestures at a maneuvers level, and low-level annotations that describe gestures at a fine-grain level. To validate this proposal, we have designed a recurrent neural network based on a bidirectional long-short term memory layer. We present results for four cross-validation experimental setups, reaching up to a 77.35% mAP.

Optimisation and Calibration of Bayesian Neural Network for Probabilistic Prediction of Biogas Performance in an Anaerobic Lagoon.

This study aims to enhance diagnostic capabilities for optimising the performance of the anaerobic sewage treatment lagoon at Melbourne Water's Western Treatment Plant (WTP) through a novel machine learning (ML)-based monitoring strategy. This strategy employs ML to make accurate probabilistic predictions of biogas performance by leveraging diverse real-life operational and inspection sensor and other measurement data for asset management, decision making, and structural health monitoring (SHM). The paper commences with data analysis and preprocessing of complex irregular datasets to facilitate efficient learning in an artificial neural network. Subsequently, a Bayesian mixture density neural network model incorporating an attention-based mechanism in bidirectional long short-term memory (BiLSTM) was developed. This probabilistic approach uses a distribution output layer based on the Gaussian mixture model and Monte Carlo (MC) dropout technique in estimating data and model uncertainties, respectively. Furthermore, systematic hyperparameter optimisation revealed that the optimised model achieved a negative log-likelihood (NLL) of 0.074, significantly outperforming other configurations. It achieved an accuracy approximately 9 times greater than the average model performance (NLL = 0.753) and 22 times greater than the worst performing model (NLL = 1.677). Key factors influencing the model's accuracy, such as the input window size and the number of hidden units in the BiLSTM layer, were identified, while the number of neurons in the fully connected layer was found to have no significant impact on accuracy. Moreover, model calibration using the expected calibration error was performed to correct the model's predictive uncertainty. The findings suggest that the inherent data significantly contribute to the overall uncertainty of the model, highlighting the need for more high-quality data to enhance learning. This study lays the groundwork for applying ML in transforming high-value assets into intelligent structures and has broader implications for ML in asset management, SHM applications, and renewable energy sectors.

Survey on Silentinterpreter : Analysis of Lip Movement and Extracting Speech using Deep Learning

Lip reading is a complex but interesting path for the growth of speech recognition algorithms. It is the ability of deciphering spoken words by evaluating visual cues from lip movements. In this study, we suggest a unique method for lip reading that converts lip motions into textual representations by using deep neural networks. Convolutional neural networks are used in the methodology to extract visual features, recurrent neural networks are used to simulate temporal context, and the Connectionist Temporal Classification loss function is used to align lip features with corresponding phonemes. The study starts with a thorough investigation of data loading methods, which include alignment extraction and video preparation. A well selected dataset with video clips and matching phonetic alignments is presented. We select relevant face regions, convert frames to grayscale, then standardize the resulting data so that it can be fed into a neural network. The neural network architecture is presented in depth, displaying a series of bidirectional LSTM layers for temporal context understanding after 3D convolutional layers for spatial feature extraction. Careful consideration of input shapes, layer combinations, and parameter selections forms the foundation of the model's design. To train the model, we align predicted phoneme sequences with ground truth alignments using the CTC loss. Dynamic learning rate scheduling and a unique callback mechanism for training visualization of predictions are integrated into the training process. After training on a sizable dataset, the model exhibits remarkable convergence and proves its capacity to understand intricate temporal correlations. Through the use of both quantitative and qualitative evaluations, the results are thoroughly assessed. We visually check the model's lip reading abilities and assess its performance using common speech recognition criteria. It is explored how different model topologies and hyperparameters affect performance, offering guidance for future research. The trained model is tested on external video samples to show off its practical application. Its accuracy and resilience in lip-reading spoken phrases are demonstrated. By providing a deep learning framework for precise and effective speech recognition, this research adds to the rapidly changing field of lip reading devices. The results offer opportunities for additional development and implementation in various fields, such as assistive technologies, audio-visual communication systems, and human-computer interaction.

MF-MNER: Multi-models Fusion for MNER in Chinese Clinical Electronic Medical Records.

To address the problem of poor entity recognition performance caused by the lack of Chinese annotation in clinical electronic medical records, this paper proposes a multi-medical entity recognition method F-MNER using a fusion technique combining BART, Bi-LSTM, and CRF. First, after cleaning, encoding, and segmenting the electronic medical records, the obtained semantic representations are dynamically fused using a bidirectional autoregressive transformer (BART) model. Then, sequential information is captured using a bidirectional long short-term memory (Bi-LSTM) network. Finally, the conditional random field (CRF) is used to decode and output multi-task entity recognition. Experiments are performed on the CCKS2019 dataset, with micro avg Precision, macro avg Recall, weighted avg Precision reaching 0.880, 0.887, and 0.883, and micro avg F1-score, macro avg F1-score, weighted avg F1-score reaching 0.875, 0.876, and 0.876 respectively. Compared with existing models, our method outperforms the existing literature in three evaluation metrics (micro average, macro average, weighted average) under the same dataset conditions. In the case of weighted average, the Precision, Recall, and F1-score are 19.64%, 15.67%, and 17.58% higher than the existing BERT-BiLSTM-CRF model respectively. Experiments are performed on the actual clinical dataset with our MF-MNER, the Precision, Recall, and F1-score are 0.638, 0.825, and 0.719 under the micro-avg evaluation mechanism. The Precision, Recall, and F1-score are 0.685, 0.800, and 0.733 under the macro-avg evaluation mechanism. The Precision, Recall, and F1-score are 0.647, 0.825, and 0.722 under the weighted avg evaluation mechanism. The above results show that our method MF-MNER can integrate the advantages of BART, Bi-LSTM, and CRF layers, significantly improving the performance of downstream named entity recognition tasks with a small amount of annotation, and achieving excellent performance in terms of recall score, which has certain practical significance. Source code and datasets to reproduce the results in this paper are available at https://github.com/xfwang1969/MF-MNER .

Shear wave velocity prediction based on 1DCNN-BiLSTM network with attention mechanism

The Shear wave (S-wave) velocity is an essential parameter in reservoir characterization and evaluation, fluid identification, and prestack inversion. However, the cost of obtaining S-wave velocities directly from dipole acoustic logging is relatively high. At the same time, conventional data-driven S-wave velocity prediction methods exhibit several limitations, such as poor accuracy and generalization of empirical formulas, inadequate exploration of logging curve patterns of traditional fully connected neural networks, and gradient explosion and gradient vanishing problems of recurrent neural networks (RNNs). In this study, we present a reliable and low-cost deep learning (DL) approach for S-wave velocity prediction from real logging data to facilitate the solution of these problems. We designed a new network sensitive to depth sequence logging data using conventional neural networks. The new network is composed of one-dimensional (1D) convolutional, bidirectional long short-term memory (BiLSTM), attention, and fully connected layers. First, the network extracts the local features of the logging curves using a 1D convolutional layer, and then extracts the long-term sequence features of the logging curves using the BiLSTM layer, while adding an attention layer behind the BiLSTM network to further highlight the features that are more significant for S-wave velocity prediction and minimize the influence of other features to improve the accuracy of S-wave velocity prediction. Afterward, the nonlinear mapping relationship between logging data and S-wave velocity is established using several fully connected layers. We applied the new network to real field data and compared its performance with three traditional methods, including a long short-term memory (LSTM) network, a back-propagation neural network (BPNN), and an empirical formula. The performance of the four methods was quantified in terms of their coefficient of determination (R2), root mean square error (RMSE), and mean absolute error (MAE). The new network exhibited better performance and generalization ability, with R2 greater than 0.95 (0.9546, 0.9752, and 0.9680, respectively), RMSE less than 57 m/s (56.29, 23.18, and 30.17 m/s, respectively), and MAE less than 35 m/s (34.68, 16.49, and 21.47 m/s, respectively) for the three wells. The test results demonstrate the efficacy of the proposed approach, which has the potential to be widely applied in real areas where S-wave velocity logging data are not available. Furthermore, the findings of this study can help for a better understanding of the superiority of deep learning schemes and attention mechanisms for logging parameter prediction.

Contextualized Satire Detection in Short Texts Using Deep Learning Techniques

Satire is prominent in user-generated content on various online platforms in the form of satirical news, customer reviews, blogs, articles, and short messages that are typically of an informal nature. As satire is also used to disseminate false information on the Internet, its computational detection has become a well-known issue. Existing work focuses primarily on formal document- or sentence-level textual data, whereas informal short texts have gotten less attention for satire detection. This paper presents a new model called BiLSTM self-attention (BiSAT) for detecting satire in informal short texts. It consists of various components such as input, embedding, self-attention, and two bi-directional long short-term memory (BiLSTM) layers for learning crucial contextual information pertaining to the satire present in the texts. The input layer uses the text as input to create an input vector, which is then given to the embedding layer to create the appropriate numeric vector. The output of the embedding layer is passed on to the first BiLSTM layer, which extracts contextual information-based sequences in the opposite direction. Between the first and second BiLSTM layers, a self-attention layer is employed to draw attention to the important satirical information that is acquired by the hidden layer of the first BiLSTM. The BiSAT model also takes a classic feature engineering approach, employing a 13-dimensional auxiliary feature vector comprised of features from four separate feature categories: sentiment, punctuation, hyperbole, and affective. The proposed BiSAT model is empirically evaluated on two benchmark datasets and a newly created dataset called Satire-280. It outperforms existing research and baseline methods by a significant margin. The Satire-280 dataset along with code can be downloaded from GitHub repository: https://github.com/Ashraf-Kamal/Satire-Detection.

Concrete acoustic emission signal augmentation method based on generative adversarial networks

With the development of concrete structure health monitoring (SHM), the utilization of acoustic emission (AE) as an effective non-destructive testing method has attracted increasing attention among researchers. However, the acquisition of concrete AE signals in real-world scenarios is often limited, which poses challenges to the training as well as the practical application of deep learning (DL) models. To address this problem, a concrete acoustic emission signal augmentation method based on generative adversarial networks (GAN) is proposed in this paper. This method features a unique architecture where both the generator and discriminator are built upon a Bidirectional Long Short-Term Memory (BiLSTM) network. Additionally, the generator integrates an attention mechanism between the BiLSTM layers, enhancing its ability to capture the temporal dynamics of AE signals. To improve the local details of the generated signals, a temporal difference mean square error (TD-MSE) regularization is incorporated into the loss function. Experimental results demonstrate that the method not only substantially increases the amount of data available but also significantly enhances the model’s generalization capabilities and classification accuracy.

Movie sentiment analysis based on Long Short-Term Memory Network

An important task in the study of Natural Language Processing (NLP) is the analysis of movie reviews. It finishes the task of classifying movie review texts into sentiment, such as positive, negative or neutral sentiment. Previous works mainly follow the pipeline of LSTM (Long Short-Term Memory Network). The network model is a variant of Recurrent Neural Network (RNN) and particularly suitable for processing natural language texts. Though existing LSTM-based works have improved the performance significantly, we argue that most of them deal with the problem of analyzing the sentiment of movie reviews while ignore analyze the model performance in different application scenarios, such as different lengths of the reviews and the frequency of sentiment adverbs in the reviews. To alleviate the above issue, in this paper, we constructed a simple LSTM model containing an embedding layer, a batch normalization layer, a dropout layer, a one-dimensional convolutional layer, a maximal pooling layer, a bi-directional LSTM layer and a fully connected layer. We used the existing IMDB movie review dataset to train the model, and selected two research scenarios of movie review length and frequency of occurrence of sentiment adverbs to test the model, respectively. From the experimental results, we proposed a model for the scenarios in which the LSTM model handles the problem of sentiment analysis with respect to the dataset construction, model stability and generalization ability, text fragment processing, data preprocessing and feature extraction, model optimization and improvement.

Novel Approach to Intrusion Detection: Introducing GAN-MSCNN-BILSTM with LIME Predictions

This paper introduces an innovative intrusion detection system that harnesses Generative Adversarial Networks (GANs), Multi-Scale Convolutional Neural Networks (MSCNNs), and Bidirectional Long Short-Term Memory (BiLSTM) networks, supplemented by Local Interpretable Model-Agnostic Explanations (LIME) for interpretability. Employing a GAN, the system generates realistic network traffic data, encompassing both normal and attack patterns. This synthesized data is then fed into an MSCNN-BiLSTM architecture for intrusion detection. The MSCNN layer extracts features from the network traffic data at different scales, while the BiLSTM layer captures temporal dependencies within the traffic sequences. Integration of LIME allows for explaining the model's decisions. Evaluation on the Hogzilla dataset, a standard benchmark, showcases an impressive accuracy of 99.16% for multi-class classification and 99.10% for binary classification, while ensuring interpretability through LIME. This fusion of deep learning and interpretability presents a promising avenue for enhancing intrusion detection systems by improving transparency and decision support in network security.

MFOA-Bi-LSTM: An optimized bidirectional long short-term memory model for short-term traffic flow prediction

Within Intelligent Transportation Systems (ITSs), modeling of traffic flow assumes a pivotal role, as it is crucial for alleviating traffic congestion and reducing carbon emissions. Developing a reliable and sustainable model for predicting traffic flow is highly challenging due to the fluctuations and nonlinear characteristics of traffic flow. In addition to being a practical problem, accurately forecasting traffic flow provides substantial challenges to the researchers working in this area. The aim of this study is to develop a complex framework for forecasting short-term traffic flow with the purpose of improving prediction accuracy. In order to attain this goal, a Bidirectional Long Short-Term Memory (Bi-LSTM) model with a Modified Firefly Optimization Algorithm (MFOA) known as MFOA-Bi-LSTM has been proposed. The MFOA optimization technique has been utilized to optimize the hyperparameters of the deep learning models. Moreover, the optimization of the Bi-LSTM network prediction model is achieved through the application of the MFOA technique, which is renowned for its ability to converge quickly, maintain robustness and perform extensive global searches. To better capture complex patterns in the input data and increase prediction accuracy, model’s architecture makes use of the bidirectional capabilities of Bi-LSTM layers. This combination of Bi-LSTM and FOA is proposed for optimizing complex sequence modeling tasks where selecting the right hyperparameters for the model is challenging and it allows for automated hyperparameter tuning and often leads to improved model performance. In terms of performance metrics including Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and Correlation Coefficient (r), the proposed model significantly outperforms the other selected models and study findings confirm its appropriateness for the predictive task.

A Synergistic Bidirectional LSTM and N-gram Multi-channel CNN Approach Based on BERT and FastText for Arabic Event Identification

Event extraction from texts continues to pose a challenge for many NLP systems. This article presents a novel neural network architecture that can extract and classify events from Arabic sentences. The model combines word representations and Part-Of-Speech (POS) tags and uses a bidirectional LSTM layer and a dual combined convolutional neural network. The first layer of the network focuses on sentence representations, while the second layer focuses on POS representations. The model takes advantage of both N-gram character features from FastText and contextual representations from bidirectional encoder representations from transformers. This combination proves to be successful, as evidenced by the good results obtained from evaluating the model on the Arabic TimeML corpus. Our results show that combining both contextual and N-gram representations outperforms the traditional skip-gram model.

Initial condition based real time classification of power quality disturbance using deep convolution neural network with bidirectional long short‐term memory

AbstractThe accurate classification of power quality disturbances (PQDs) is crucial for advancing real‐time monitoring and classification systems within the modern power grid. The proposed system must ensure dependable, safeguarded, and stable operating conditions amidst diverse power quality issues. This paper presents an approach to classifying power quality disturbances using a deep learning model that synergizes deep convolutional neural networks (DCNN) and Bidirectional Long Short‐Term Memory (BiLSTM). This amalgamation effectively extracts and classifies disturbance signals in real time, grounded on noise levels. The initial feature extraction from the signal is accomplished through a time‐frequency matrix. Subsequently, secondary extraction employs the BiLSTM layer to intricately and significantly classify disturbances in the power signal. This aids in transforming high‐dimensional matrices into a reduced set for enhanced performance. The detailed classification is facilitated by the softmax layer. The simulation results support the power quality evaluations under varied constraints and underscore the substantial classification of power quality disturbances through the DCNN‐BiLSTM algorithm, in comparison to alternative classification algorithms in terms of computational speed and accuracy.

Advanced detection of cardiac arrhythmias using a three-stage CBD filter and a multi-scale approach in a combined deep learning model

In this paper, we present a new method based on deep learning for the detection of cardiac arrhythmias. The proposed method is based on three fundamental components: an inception-ResNet structure, bidirectional LSTM layers and a multi-scale signal analysis technique. To improve processing of the MIT-BIH arrhythmia dataset, we have developed a three-stage CBD filter that successively incorporates Chebyshev, Butterworth and Daubechies filters. This filter is designed meticulously to denoise the ECG signal while preserving its relevant features. By combining this filter with our deep learning model and multi-scale approach, we have significantly improved the quality of the ECG signal. Such integration not only enhances the accuracy and reliability of arrhythmia detection, but also contributes to the identification of complex patterns in various temporal and spatial dimensions. In terms of performance, our model is remarkably robust, with 98.94% accuracy and 97.20% precision.

Towards job screening and personality traits estimation from video transcriptions

In recent years, natural language processing (NLP) has gained new territory beyond its traditional use in text mining applications. This paper shows the effectiveness of NLP techniques in assessing the apparent human personality from his/her video transcript, building a bridge between NLP and computer vision-based reasoning. In this paper, a new deep learning model using attention mechanism and bidirectional LSTM layers for estimating the Big-five personality traits is provided and then tested on ChatLearn video dataset. The robustness of the approach is then tested by generalizing the method to two other datasets (B5 corpus and Mypersonality datasets). Several empirical evaluations taking into account the various inputs of the data processing pipeline have been performed to yield optimal model parameters. The developed model is tested on the APA’2016 competition dataset from Chalearn V2 Challenge Workshop for both Big-five personality trait estimation and interview score estimation. We achieved an average of 89.27% in personality trait recognition rate and 89.16% score in the Job Interview challenge. Similar trends of high accuracy score estimation is held for B5-corpus and MyPersonality datasets as well. The results outperformed several state-of-art approaches, demonstrating our approach’s feasibility in extending the computer vision approach of first impression personality to natural language processing. This opens up a new direction in multimedia analysis.

Named Entity Recognition of Electronic Medical Records based on BERT-BiLSTM-Biaffine Model

In the case of the specific task of identifying named entities within electronic medical record, it is hard to determine the boundary of nested entities, and existing NER systems have insufficient decoding performance. Based on the pre training model BERT, this paper introduces a novel network structure called Biaffine Layer using a bidirectional LSTM layer. The network uses a dual affine attention mechanism for semantic information learning, which can better interact with the semantic information of entity heads and entity tails, thereby achieving better recognition results for entities. Due to the sparsity of named entity datasets and the uneven distribution of entity categories, traditional binary cross entropy loss functions require multiple rounds of training to decode entities. In this paper, we have modified the binary cross entropy loss function to make the proposed model faster decode the entities that need to be identified. The model performs well, according to the experimental findings. The approach suggested in this paper offers a fresh approach to the NER issue raised by electronic medical records, and it is anticipated to considerably boost the effectiveness and caliber of clinical medical research.

Attention-assisted hybrid 1D CNN-BiLSTM model for predicting electric field induced by transcranial magnetic stimulation coil

Deep learning-based models such as deep neural network (DNN) and convolutional neural network (CNN) have recently been established as state-of-the-art for enumerating electric fields from transcranial magnetic stimulation coil. One of the main challenges related to this electric field enumeration is the prediction time and accuracy. Despite the low computational cost, the performance of the existing prediction models for electric field enumeration is quite inefficient. This study proposes a 1D CNN-based bi-directional long short-term memory (BiLSTM) model with an attention mechanism to predict electric field induced by a transcranial magnetic stimulation coil. The model employs three consecutive 1D CNN layers followed by the BiLSTM layer for extracting deep features. After that, the weights of the deep features are redistributed and integrated by the attention mechanism and a fully connected layer is utilized for the prediction. For the prediction purpose, six input features including coil turns of single wing, coil thickness, coil diameter, distance between two wings, distance between head and coil position, and angle between two wings of coil are mapped with the output of the electric field. The performance evaluation is conducted based on four verification metrics (e.g. R2, MSE, MAE, and RMSE) between the simulated data and predicted data. The results indicate that the proposed model outperforms existing DNN and CNN models in predicting the induced electrical field with R2 = 0.9992, MSE = 0.0005, MAE = 0.0188, and RMSE = 0.0228 in the testing stage.

Multi-Step Ahead Forecasting of the Energy Consumed by the Residential and Commercial Sectors in the United States Based on a Hybrid CNN-BiLSTM Model

COVID-19 has continuously influenced energy security and caused an enormous impact on human life and social activities due to the stay-at-home orders. After the Omicron wave, the economy and the energy system are gradually recovering, but uncertainty remains due to the virus mutations that could arise. Accurate forecasting of the energy consumed by the residential and commercial sectors is challenging for efficient emergency management and policy-making. Affected by geographical location and long-term evolution, the time series of the energy consumed by the residential and commercial sectors has prominent temporal and spatial characteristics. A hybrid model (CNN-BiLSTM) based on a convolution neural network (CNN) and bidirectional long short-term memory (BiLSTM) is proposed to extract the time series features, where the spatial features of the time series are captured by the CNN layer, and the temporal features are extracted by the BiLSTM layer. Then, the recursive multi-step ahead forecasting strategy is designed for multi-step ahead forecasting, and the grid search is employed to tune the model hyperparameters. Four cases of 24-step ahead forecasting of the energy consumed by the residential and commercial sectors in the United States are given to evaluate the performance of the proposed model, in comparison with 4 deep learning models and 6 popular machine learning models based on 12 evaluation metrics. Results show that CNN-BiLSTM outperforms all other models in four cases, with MAPEs ranging from 4.0034% to 5.4774%, improved from 0.1252% to 49.1410%, compared with other models, which is also about 5 times lower than that of the CNN and 5.9559% lower than the BiLSTM on average. It is evident that the proposed CNN-BiLSTM has improved the prediction accuracy of the CNN and BiLSTM and has great potential in forecasting the energy consumed by the residential and commercial sectors.

Deep belief improved bidirectional LSTM for multivariate time series forecasting.

Multivariate time series (MTS) play essential roles in daily life because most real-world time series datasets are multivariate and rich in time-dependent information. Traditional forecasting methods for MTS are time-consuming and filled with complicated limitations. One efficient method being explored within the dynamical systems is the extended short-term memory networks (LSTMs). However, existing MTS models only partially use the hidden spatial relationship as effectively as LSTMs. Shallow LSTMs are inadequate in extracting features from high-dimensional MTS; however, the multilayer bidirectional LSTM (BiLSTM) can learn more MTS features in both directions. This study tries to generate a novel and improved BiLSTM network (DBI-BiLSTM) based on a deep belief network (DBN), bidirectional propagation technique, and a chained structure. The deep structures are constructed by a DBN layer and multiple stacked BiLSTM layers, which increase the feature representation of DBI-BiLSTM and allow for the model to further learn the extended features in two directions. First, the input is processed by DBN to obtain comprehensive features. Then, the known features, divided into clusters based on a global sensitivity analysis method, are used as the inputs of every BiLSTM layer. Meanwhile, the previous outputs of the shallow layer are combined with the clustered features to reconstitute new input signals for the next deep layer. Four experimental real-world time series datasets illustrate our one-step-ahead prediction performance. The simulating results confirm that the DBI-BiLSTM not only outperforms the traditional shallow artificial neural networks (ANNs), deep LSTMs, and some recently improved LSTMs, but also learns more features of the MTS data. As compared with conventional LSTM, the percentage improvement of DBI-BiLSTM on the four MTS datasets is 85.41, 75.47, 61.66 and 30.72%, respectively.