Search

• We propose a hyper-clustering (hyper-AP clustering) approach to capture the migration trend between individuals and clusters in intelligent transportation systems. • We use the hyper-clustering enhanced spatio-temporal deep learning model such as ST-ResNet and ST-3DNet, and we show that our approach can significantly improve the prediction accuracy of both deep learning methods. • Compared with state-of-the-art methods, the improvement of the prediction performance of our model is 10.9 Bike traffic prediction is essential to operations management in bike-sharing systems (BSS). The spatio-temporal distribution of shared bikes is imbalanced; therefore, it is necessary to predict the number of bikes that are available at each station to facilitate operations. Many researchers have developed deep neural network methods to improve predictive performance. Predicting bike demand from a higher level by clustering mobility trends has been shown to increase prediction accuracy. However, complex iterative optimization problems need to be solved to stabilize the trend iteration, and such solutions are missing in previous works. In this paper, we propose a novel hyper-clustering approach that enhances a spatio-temporal deep neural network for traffic prediction in bike-sharing systems. In particular, our hyper-clustering approach captures mobility trends among individuals and clusters. Experimental results show that the designed hyper-clustering-enhanced spatio-temporal deep neural network model can predict the number of available bikes more accurately. Compared with ten state-of-the-art benchmark methods, our model demonstrates better performance in reducing prediction error by 10.9%–43.1%. [ABSTRACT FROM AUTHOR]

• We propose SPLNet for multi-step regional wind speed prediction, which is based on multi-time Seq2one learning. It can effectively build the long-term temporal and the global spatial dependencies by a new STD-Atten unit. • SPLNet adopts a time-variant structure to capture the fluctuate, intermittent and chaotic nature of wind. Moreover, a new step-separated training strategy is proposed to stabilize model training and enhance the performance. • We have conducted extensive experiments on three realistic datasets. The experimental results show that the proposed SPLNet outperforms state-of-the-art methods. Moreover, ablation study results demonstrate the effectiveness of designed components in SPLNet. Although regional wind speed prediction is of great value in the development of wind power, it is rarely studied in the past years. To fill this gap, we introduce a deep learning method for multi-step regional wind speed prediction, which aims to model the spatiotemporal dynamics of wind speed in a given region and make predictions across time steps. There are several important challenges for it: (1) how to simultaneously build long-term temporal dependencies and long-range spatial interactions of wind speed; (2) how to capture the fluctuant, intermittent, and chaotic nature of wind speed. In this paper, we propose a novel framework named Sequence-to-one Predictive Learning Net (SPLNet) for multi-step regional wind speed prediction. To effectively utilize historical information and build the long-term temporal dependencies, SPLNet regards multi-step prediction as multi-time sequence-to-one (Seq2one) learning, where each prediction is made based on a previous historical sequence. Besides, a new spatiotemporal dynamics attention (STD-Atten) unit is developed as the key component of SPLNet, which can realize Seq2one learning and capture long-range spatial dependencies. Moreover, to better model the non-stationary and violent nature of wind speed, SPLNet adopts a time-variant structure, where different STD-Atten units are used for predicting wind speed at different time steps. To stabilize model training and enhance prediction performance, a step-separated training procedure is put forward. Comprehensive experiments have been conducted on three real-world datasets, where regional wind speed is illustrated as images. Experimental results demonstrate the effectiveness and superiority of the proposed method over state-of-the-art techniques. [ABSTRACT FROM AUTHOR]

Acute hypotensive episode (AHE) is a common serious postoperative complication in ICU, which may raise multiple system failure (especially of cardiac and respiratory kinds), and even cause death. Timely and effective clinical intervention is obviously vital to the saving of patients. AHE detection involves physiological time-series monitoring, processing and prediction technologies, which can offer insights to neuroscientists, biologists, and even provide support for clinicians. This paper presents a hybrid artificial intelligence model combined with CEEMDAN (complete ensemble empirical mode decomposition with adaptive noise, a typical method for physiological signal decomposition), deep learning, multiple gene expression programming and fuzzy expert system for AHE detection. In this paper, the physiological signal is selected from a benchmark dataset, for example MIMIC-II (Multiparameter Intelligent Monitoring in Intensive Care II), which collects large scale real patients' data for clinical research. In the hybrid model, a typical signal decomposition method is employed for AHE signal processing, and an autoencoder based deep neural network is established for feature extraction. Finally, a reliable and explainable classifier is presented by fusing gene expression programming and the fuzzy method. Experimental results based on real data set demonstrate that the proposed method outperforms state-of-the-art AHE detection methods by achieving the prediction accuracy of 88.14% in 2866 records. [ABSTRACT FROM AUTHOR]

Time-series data are widespread in real-world industrial scenarios. To recover and infer missing information in real-world applications, the problem of time-series prediction has been widely studied as a classical research topic in data mining. Deep learning architectures have been viewed as next-generation time-series prediction models. However, recent studies have shown that deep learning models are vulnerable to adversarial attacks. In this study, we prospectively examine the problem of time-series prediction adversarial attacks and propose an attack strategy for generating an adversarial time series by adding malicious perturbations to the original time series to deteriorate the performance of time-series prediction models. Specifically, a perturbation-based adversarial example generation algorithm is proposed using the gradient information of the prediction model. In practice, unlike the imperceptibility to humans in the field of image processing, time-series data are more sensitive to abnormal perturbations and there are more stringent requirements regarding the amount of perturbations. To address this challenge, we craft an adversarial time series based on the importance measurement to slightly perturb the original data. Based on comprehensive experiments conducted on real-world time-series datasets, we verify that the proposed adversarial attack methods not only effectively fool the target time-series prediction model LSTNet, they also attack state-of-the-art CNN-, RNN-, and MHANET-based models. Meanwhile, the results show that the proposed methods achieve a good transferability. That is, the adversarial examples generated for a specific prediction model can significantly affect the performance of the other methods. Moreover, through a comparison with existing adversarial attack approaches, we can see that much smaller perturbations are sufficient for the proposed importance-measurement based adversarial attack method. The methods described in this paper are significant in understanding the impact of adversarial attacks on a time-series prediction and promoting the robustness of such prediction technologies. [ABSTRACT FROM AUTHOR]

This paper addresses the biological plausibility of both backpropagation (BP) and contrastive Hebbian learning (CHL) used in the Boltzmann machines. The main claim of this paper is that CHL is a general learning algorithm that can be used to steer feedforward networks toward desirable outcomes, and steer them away from undesirable outcomes without any need for the specialized feedback circuit of BP or the symmetric connections used by the Boltzmann machines. After adding perturbations during the learning phase to all the neurons in the network, multiple feedforward outcomes are classified into Hebbian and anti-Hebbian sets based on the network predictions. The algorithm is applied to networks when optimizing a loss objective where BP excels and is also applied to networks with stochastic binary outputs where BP cannot be easily applied. The power of the proposed algorithm lies in its simplicity where both learning and gradient estimation through stochastic binary activations are combined into a single local Hebbian rule. We will also show that both Hebbian and anti-Hebbian correlations are evaluated from the readily available signals that are fundamentally different from CHL used in the Boltzmann machines. We will demonstrate that the new learning paradigm where Hebbian/anti-Hebbian correlations are based on correct/incorrect predictions is a powerful concept that separates this paper from other biologically inspired learning algorithms. [ABSTRACT FROM AUTHOR]

In this paper, we introduce ProNet, an novel deep learning approach designed for multi-horizon time series forecasting, adaptively blending autoregressive (AR) and non-autoregressive (NAR) strategies. Our method involves dividing the forecasting horizon into segments, predicting the most crucial steps in each segment non-autoregressively, and the remaining steps autoregressively. The segmentation process relies on latent variables, which effectively capture the significance of individual time steps through variational inference. In comparison to AR models, ProNet showcases remarkable advantages, requiring fewer AR iterations, resulting in faster prediction speed, and mitigating error accumulation. On the other hand, when compared to NAR models, ProNet takes into account the interdependency of predictions in the output space, leading to improved forecasting accuracy. Our comprehensive evaluation, encompassing four large datasets, and an ablation study, demonstrate the effectiveness of ProNet, highlighting its superior performance in terms of accuracy and prediction speed, outperforming state-of-the-art AR and NAR forecasting models. • Forecasting of electricity energy consumption and solar energy generation. • Informer-based model marrying forecasting horizon segmentation and variational inference techniques. • Comparing SARIMAX, DeepAR, DeepSSM, LogTrans, N-BEATS, and Informer models forecasting ability. • Performance evaluation based on four real-world datasets. [ABSTRACT FROM AUTHOR]

• Optimized deep learning, statistical and hybrid models are proposed for deterministic (point) and probabilistic (interval) forecasting of crude oil price (COP). • A DE-DL method employing differential evolution algorithm is proposed to optimize the architecture and other hyper-parameters of deep learning models for a problem. • Recurrence plot is drawn, recurrence quantification analysis is carried out and uncertainty analysis of prediction error is performed using Gaussian and t-Location Scale distributions. • Extensive non-parametric statistical analysis on the obtained results are carried out to rank the optimized models and draw decisive conclusions relating to deterministic and probabilistic forecasting of COP. • Eight research questions related to deterministic and probabilistic forecasting of COP are addressed. In this paper, individual and hybrid methods are proposed employing optimized statistical and deep learning (DL) models for deterministic (point) and probabilistic (interval) forecasting of crude oil price time series. The statistical models are optimized using the Forecast package of R. To enhance the performance of DL models, a novel pruning DE-DL method is proposed, which employs the differential evolution (DE) algorithm to optimize architecture and continuous and discrete-valued hyper-parameters. The proposed DE-DL method is so generic that it can be applied to optimize different DL models for any supervised learning problem. Five DL models (LSTM, BiLSTM, GRU, CNN, and ConvLSTM) are optimized for forecasting monthly crude oil prices and hybridized with an optimized ARIMA model for developing optimized additive and multiplicative hybrid forecasting models. The effectiveness of the proposed methods is evaluated through deterministic and probabilistic forecasting measures, comparing the results with six optimized statistical models, thirteen machine learning models, five optimized DL models, and ten optimized hybrid models. It is observed from the simulation results that the proposed optimized Additive-ARIMA-GRU hybrid model provides statistically superior forecasts, and the t Location Scale distribution is more suitable than the Gaussian distribution for computing reliable prediction intervals with different significance levels. [ABSTRACT FROM AUTHOR]

This paper introduces deep learning-based forecasting models for the continuous prediction of the aggregated production generated by CSP plants in Spain. These models use as inputs the expected top of atmosphere irradiance values and available weather conditions forecasts for the locations where the main CSP power plants are installed. The performances of the forecast models are analysed and compared by means of the most extended metrics in the literature for a whole year of CSP energy production. [ABSTRACT FROM AUTHOR]

Deep learning (DL) techniques have shown good potential in building favorable predictive models in e-learning environments. It is important to build a DL-based, fine-grained student performance prediction model to predict students' outcomes and learning status at every stage of their course, rather than merely predicting the students' final score and drop-out rate, which is referred to as coarse-grained performance. In this paper, we tackle the problem of fine-grained student performance prediction in an online course using DL-based long sequence generation by formulating the students' feature as a matrix where elements in certain parts are missing. We propose an adaptive sparse self-attention network to generate the missing values and simultaneously predict the fine-grained performance. The matrix representation of the features facilitates position-wise feature selection, which helps to find the most correlated components from the past learning stage, leading to an embedding with both the original features and their spatial relationships. We build a deep neural network model with several sparse self-attention layers stacked together to achieve sequence generation. Experimental studies on three real-world datasets from different e-learning platforms demonstrate the effectiveness and advantages of our proposed method. [ABSTRACT FROM AUTHOR]

Fuzzy Cognitive Maps (FCMs) have become a relevant technique for modeling and forecasting time series due to their advantages in dealing with uncertainty and simulating the dynamics of complex systems. Although numerous univariate and multivariate FCM-based forecasting models have been presented in the literature, one of the still open questions is how to enable FCMs to forecast multivariate time series for multiple-input, multiple-output (MIMO) systems with an efficient learning mechanism. from a computational point of view. This paper suggests a randomized MIMO FCM-based forecasting approach called MO-RHFCM to predict low-dimensional multivariate time series. More specifically, MO-RHFCM is a hybrid model merging the concepts of multivariate fuzzy time series, high order FCM (HFCM), and Echo State Networks (ESN). The structure of MO-RHFCM consists of three layers: input layer, reservoir (internal) layer, and output layer. Only the output layer is trainable using the Least Squares minimization algorithm; hence training the proposed MO-RHFCM method is fast and simple. The weights inside each sub-reservoir are selected randomly and remain fixed during the training process. The obtained results indicate the efficacy and validity of the proposed MO-RHFCM technique compared with some machine learning and deep learning baseline models. • Introducing a new randomized multivariate FCM model as a reservoir computing. • The first introduction of FCM-based forecasting method for multiple outputs systems. • The superior performance of the proposed model in comparison to other baselines. • Less complexity of our proposed model compared to baseline deep learning models. [ABSTRACT FROM AUTHOR]

• A novel stacked isomorphic autoencoder (SIAE) is proposed for soft sensing development. • SIAE seeks to extract abstract features at each layer from its previous one by stacking hierarchical isomorphic autoencoders, which reconstructs the same raw data as well as possible at each layer. • SIAE can better describe the complex data patterns of the observed raw data and obtain good features. • The effectiveness and flexibility of the proposed method are validated on an industrial sulfur recovery unit. Deep learning is an important and effective tool for process soft sensor modeling in industrial artificial intelligence. Traditional deep learning methods like stacked autoencoder (SAE) usually learn high-level features from their low-level ones progressively by minimizing the reconstruction error for the inputs at each layer. However, the reconstruction cannot be exactly accurate. There is loss cumulation of raw data information from the lowest to the highest levels in SAE. To deal with this problem, a novel deep stacked isomorphic autoencoder (SIAE) is proposed to obtain better feature representation for raw input data in this paper. Different from the original SAE, SIAE aims to extract abstract features at each layer from its previous one by stacking hierarchical isomorphic autoencoders (IAE), in which each IAE reconstructs the same raw input data as well as possible. Thus, SIAE can better describe the complex data patterns and obtain good features for the raw data. Then, SIAE is used to construct soft sensor model for quality prediction. The application on an industrial sulfur recovery unit shows that SIAE can improve the prediction performance for the quality variable. [ABSTRACT FROM AUTHOR]

In the refining process, Fluid Catalytic Cracking Unit (FCCU) end-point quality prediction plays an important role in real-time quality monitoring, optimization and control. Due to the process uncertainties (such as changes of raw materials, abrasion of mechanic components, catalyst deactivation, changes in the external environment, etc), data distribution of process data in FCCU is time-varying, which leads to accuracy degradation of the quality prediction model. Therefore, to deal with the time varying characteristic of process data and avoid accuracy degradation of the quality prediction model, real-time processing should be considered. In this paper, an enhancing incremental deep learning approach is proposed for the online quality prediction of the absorption-stabilization system in FCCU. First, the offline model is built by Stacked Auto Encoder-Deep Neural Network (SAE-DNN). To determine whether the data distribution has changed and model modification is needed, a concept drift detector is proposed for the regression problem by defining an error bound in the Statistical Test of Equal Proportions (STEPD). If the model modification is needed, then the top layer of the offline SAE-DNN model is expanded by Random Vector Functional Link (RVFL) structure, and the parameters in the expansion layer is dynamically assigned by the new coming data with the group lasso regularization and the L2 regularization. The proposed approach is validated by predicting the Saturated Vapor Pressure (SVP) of stabilized gasoline in the FCCU. The experimental results show that the proposed approach can deal with the time-varying characteristic of process data and avoid accuracy degradation under process uncertainties. [ABSTRACT FROM AUTHOR]

Many severe epidemics and pandemics have hit human civilizations throughout history. The recent Sever Actuate Respiratory disease SARS-CoV-2 known as COVID-19 became a global disease and is still growing around the globe. It has severely affected the world's economy and ways of life. It necessitates predicting the spread in advance and considering various control policies to avoid the country's complete closure. In this paper, we propose deep learning-based stacked Bi-directional long short-term memory (Stacked Bi-LSTM) network that forecasts COVID-19 more accurately for the country of South Korea. The paper's main objectives are to present a lightweight, accurate, and optimized model to predict the spread considering restriction policies such as school closure, workspace closing, and the canceling of public events. Based on the fourteen parameters (including control policies), we predict and forecast the future value of the number of positive, dead, recovered, and quarantined cases. In this paper, we use the dataset of South Korea comprised of several control policies implemented for minimizing the spread of COVID-19. We compare the performance of the stacked Bi-LSTM with the traditional time-series models and LSTM model using the performance metrics mean absolute error (MAE), mean absolute percentage error (MAPE), and root mean square error (RMSE). Moreover, we study the impact of control policies on forecasting accuracy. We further study the impact of changing the Bi-LSTM default activation functions Tanh with ReLU on forecasting accuracy. The research provides insight to policymakers to optimize the pooling of resources more optimally on the correct date and time prior to the event and to control the spread by employing various strategies in the meantime. • Proposed Policies based COVID-19 Spread Forecasting Model to examine the impact of spread in South Korea. • Proposed Stacked Bi-LSTM Novel Forecasting Model for COVID-19 Spread Control. • A comparison between Stacked Bi-LSTM and LSTM is presented. • Examined individual and collective 10 policies such as school closure, restriction on travel and vice versa. • Provided the multiple combination of COVID-19 Spread control policies analysis for future Forecasting to minimize the spread. [ABSTRACT FROM AUTHOR]

Deep neural networks bring in impressive accuracy in various applications, but the success often relies on heavy network architectures. Taking well-trained heavy networks as teachers, classical teacher–student learning paradigm aims to learn a student network that is lightweight yet accurate. In this way, a portable student network with significantly fewer parameters can achieve considerable accuracy, which is comparable to that of a teacher network. However, beyond accuracy, the robustness of the learned student network against perturbation is also essential for practical uses. Existing teacher-student learning frameworks mainly focus on accuracy and compression ratios, but ignore the robustness. In this paper, we make the student network produce more confident predictions with the help of the teacher network, and analyze the lower bound of the perturbation that will destroy the confidence of the student network. Two important objectives regarding prediction scores and gradients of examples are developed to maximize this lower bound, to enhance the robustness of the student network without sacrificing the performance. Experiments on benchmark data sets demonstrate the efficiency of the proposed approach to learning robust student networks that have satisfying accuracy and compact sizes. [ABSTRACT FROM AUTHOR]

Probability density forecast offers the whole distributions of forecasting targets, which brings greater flexibility and practicability than the other probabilistic forecast models such as prediction interval (PI) and quantile forecast. However, existing density forecast models have introduced various constraints on forecasted distributions, which has limited their ability to approximate real distributions and may result in suboptimality. In this paper, a distribution-free density forecast model based on deep learning is proposed, in which the real cumulative density functions (CDFs) of forecasting target are approximated by a large-capacity positive-weighted deep neural network (NN). Benefiting from the universal approximation ability of NNs, the range of forecasted distributions has been proven to contain all the distributions with continuous CDFs, which is superior to existing models’ considering both width and accordance with reality. Three tests from different scenarios were implemented for evaluation, i.e., very-short-term wind power, wind speed, and day-ahead electricity price forecast, in which the proposed density forecast model has shown superior performance over the state of the art. [ABSTRACT FROM AUTHOR]

Time series are a kind of streaming data, which are chaotic and sequential. As real-world time series data are often not available at once and drift with time growth, Incremental Learning (IL) is well suited for Time Series Prediction (TSP). Most previous incremental TSP algorithms are limited by the assumption of data balance. However, real-world time series data are often unbalanced, with long-tailed distribution and other characteristics resulting in the failure of IL algorithms. In this paper, a balanced-driven Active Learning (AL) strategy is proposed to deal with data imbalance problems in IL processes. What's more, by integrating the advantages of Deep Learning (DL) and the Broad Learning System (BLS), a novel Deep-Broad Learning (DeepBL) network with its incremental learning algorithm is proposed. The proposed A ctive Learning-based I ncremental Deep - B road L earning (AI_DeepBL) algorithm is applied to real-world univariate and multivariate time series datasets and achieves superior performance compared with classical and state-of-the-art TSP algorithms. [ABSTRACT FROM AUTHOR]

Molecular property prediction is an important task in drug discovery, especially the characterization of relationships between molecular substructures and their property. It is usually implemented by deep learning methods. However, deep model is a black-box that would lead to the prediction process and results are unbelievable and unreliable. To address these issues, researchers have begun to work on investigating explainable or substitute models for deep model. In this paper, an explainable framework of molecular property prediction with their multi-granularity representation (MgR) is proposed to characterize the contribution of substructures to prediction, called MgRX for short ('X' comes from 'eXplainable'). Specifically, the MgRX is constructed to denote the substructures' contribution of different granularity, in which each substructure is progressively finer from top to bottom. The finest substructures' contribution to target is implemented by the deep learning model with SHAP. Besides, several experiments are executed to analyze the effectiveness of the multi-granularity framework to molecular property. Based on above discussions, we develop a basic framework of quantitative characterization on substructures' contribution to property via multi-granularity. [ABSTRACT FROM AUTHOR]

Forecasting accuracy is an important issue for successful intelligent traffic management, especially in the domain of traffic efficiency and congestion reduction. The dawning of the big data era brings opportunities to greatly improve prediction accuracy. In this paper, we propose a novel model, stacked autoencoder Levenberg-Marquardt model, which is a type of deep architecture of neural network approach aiming to improve forecasting accuracy. The proposed model is designed using the Taguchi method to develop an optimized structure and to learn traffic flow features through layer-by-layer feature granulation with a greedy layerwise unsupervised learning algorithm. It is applied to real-world data collected from the M6 freeway in the U.K. and is compared with three existing traffic predictors. To the best of our knowledge, this is the first time that an optimized structure of the traffic flow forecasting model with a deep learning approach is presented. The evaluation results demonstrate that the proposed model with an optimized structure has superior performance in traffic flow forecasting. [ABSTRACT FROM AUTHOR]

• Explicitly modeling coding features helps prediction of TIS. • Integrating coding features with consensus motifs improves prediction of TIS. • CNN is sensitive to the first reading frame. Translation initiation site (TIS) prediction is one of the most crucial subtasks for gene annotation. Many computational methods have been proposed and achieved acceptable accuracy in transcripts (e.g., cDNA, mRNA). However, the prediction of TIS at the genome level is far more challenging and the computational methods for TIS prediction in genomic sequences so far reach modest performance. Recently, we proposed a method that improves the prediction of TIS in mRNA sequences and demonstrated the significance of explicitly modeling coding features. In this paper, we extend the same results to genomic sequence and present a two-stage deep learning model for TIS prediction in genomic sequence: the first stage to extract coding contrast features around TIS by a hybrid Convolutional Neural Network-Bidirectional Recurrent Neural Network architecture (Content-RCNN), and the second stage to integrate coding contrast features around TIS with TIS sequence encoded by one-hot encoding to jointly predict TIS by a CNN (Integrated-CNN). Four-fold cross validation tests on genome-wide human and mouse datasets demonstrate that the proposed model yields an improved prediction performance of TIS over existing state-of-the-art methods. The source code and the dataset used in the paper are publicly available at: https://github.com/xdcwei/DeepTIS/. [ABSTRACT FROM AUTHOR]

To improve the prediction accuracy of chaotic time series, deep extraction of the system evolutionary patterns is a key problem in modeling. In this paper, we propose a deep learning model of automatic multi-scale feature extraction for chaotic time series prediction. A hybrid deep neural network named deep temporal-inception module and gated recurrent unit network (DTIGNet) is designed. The improved temporal-inception module stacks dilated causal convolution of different depth to increase the network adaptability to different scales and improve the network nonlinear characterization ability, and an optional 1 × 1 convolutional kernel as shared residual connection. The model is applied to the Mackey-Glass system, Rossler system, Lorenz system and sunspots time series to verify the applicability and effectiveness in chaotic time series prediction. The results show that the DTIGNet proposed has higher accuracy and better performance compared with other methods according to the 6 prediction evaluation metrics adopted. • We propose a deep learning model (DTIGNet) of automatic multi-scale feature extraction for chaotic time series prediction. • The improved temporal-inception module proposed extracts data features at different scales. • The results show that the proposed model has lower prediction errors and superior performance than existing models. [ABSTRACT FROM AUTHOR]

• Impact of covariates on COVID-19 response were tested via Granger-causality tests. • Climatic impact on COVID-19 cases-to-mortality ratios were studied in 36 countries. • Relationship equations were established using regression analysis. • Temperature data were factored-in in training forecasting models. • The trained models were used for forecasting COVID-19 cases-to-mortality ratios. There are several recent publications criticizing the failure of COVID-19 forecasting models, with swinging over predictions and underpredictions, which have made it difficult for decision and policy making. Observing the failures of several COVID-19 forecasting models and the alarming spread of the virus, we seek to use some stable response for forecasting COVID-19, viz., ratios of COVID-19 cases to mortalities, rather than COVID-19 cases or fatalities. A trend of low COVID-19 cases-to-mortality ratios calls for urgent attention: the need for vaccines, for instance. Studies have shown that there are influences of weather parameters on COVID-19; and COVID-19 may have come to stay and could manifest a seasonal outbreak profile similar to other infectious respiratory diseases. In this paper, the influences of some weather, geographical, economic and demographic covariates were evaluated on COVID-19 response based on a series of Granger-causality tests. The effect of four weather parameters, viz., temperature, rainfall, solar irradiation and relative humidity, on daily COVID-19 cases-to-mortality ratios of 36 countries from 5 continents of the world were determined through regression analysis. Regression studies show that these four weather factors impact ratios of COVID-19 cases-to-mortality differently. The most impactful factor is temperature which is positively correlated with COVID-19 cases-to-mortality responses in 24 out of 36 countries. Temperature minimally affects COVID-19 cases-to-mortality ratios in the tropical countries. The most influential weather factor – temperature – was incorporated in training random forest and deep learning models for forecasting the cases-to-mortality rate of COVID-19 in clusters of countries in the world with similar weather conditions. Evaluation of trained forecasting models incorporating temperature features show better performance compared to a similar set of models trained without temperature features. This implies that COVID-19 forecasting models will predict more accurately if temperature features are factored in, especially for temperate countries. [ABSTRACT FROM AUTHOR]

Fault prediction technology provides a way to reduce the loss caused by equipment failure. Currently, many efforts of failure prediction pay more attention to the fault trend and the remaining useful life. This paper devotes to predicting a specific failure mode of device in the future. We propose a two-step prediction method with two sub-models to predict the fault mode. The first model, named regression model, combines Attention-based-LSTM (Long Short-Term Memory) and Random Forest (RF) to predict time series trends, where wavelet packet analysis is adopted to extract elementary features of time series and form multivariable time series. Based on the status data forecasted by the first model, the second model, named classification model, combines Attention-based-LSTM and Extra-Tree (ET) to classify fault mode. The classified fault will occur in the device at some point in the future. In the regression model, Attention-based-LSTM performs deep level feature extraction of vibration sensor signals, and the result is fed to the last layer (i.e., an RF layer), which takes responsibility for predicting the future sequence. The future sequence is then provided to the classification model, where Attention-based-LSTM is used to extract implicit features, and ET classifier is used to accept these features and figure out the failure phenomenon. Our approach for predicting failure modes includes the severity and type of fault, which are rarely investigated synthetically. The experimental results of bearing data show that the sub-models can achieve high prediction accuracy and our method can predict the future failure mode with precision. [ABSTRACT FROM AUTHOR]

The coronavirus COVID-19 is affecting 213 countries and territories around the world. Iran was one of the first affected countries by this virus. Isfahan, as the third most populated province of Iran, experienced a noticeable epidemic. The prediction of epidemic size, peak value, and peak time can help policymakers in correct decisions. In this study, deep learning is selected as a powerful tool for forecasting this epidemic in Isfahan. A combination of effective Social Determinant of Health (SDH) and the occurrences of COVID-19 data are used as spatiotemporal input by using time-series information from different locations. Different models are utilized, and the best performance is found to be for a tailored type of long short-term memory (LSTM). This new method incorporates the mutual effect of all classes (confirmed/ death / recovered) in the prediction process. The future trajectory of the outbreak in Isfahan is forecasted with the proposed model. The paper demonstrates the positive effect of adding SDHs in pandemic prediction. Furthermore, the effectiveness of different SDHs is discussed, and the most effective terms are introduced. The method expresses high ability in both short- and long- term forecasting of the outbreak. The model proves that in predicting one class (like the number of confirmed cases), the effect of other accompanying numbers (like death and recovered cases) cannot be ignored. In conclusion, the superiorities of this model (particularity the long term predication ability) turn it into a reliable tool for helping the health decision-makers. [ABSTRACT FROM AUTHOR]

• A predicting model for the long-term epidemic trend of COVID-19 by using LSTM with rolling update mechanism is proposed. • The 150-days ahead epidemic trend of COVID-19 in Russia, Peru and Iran are estimated by our proposed model. • The results provide that the epidemic of Peru will end in early December. • The number of daily cases in Russia and Iran is expected to fall below 2000 and 1000 by mid-November and early December. • By introducing diffusion index, the effectiveness of preventive measures taken by the government are analyzed. The COVID-19 outbreak in late December 2019 is still spreading rapidly in many countries and regions around the world. It is thus urgent to predict the development and spread of the epidemic. In this paper, we have developed a forecasting model of COVID-19 by using a deep learning method with rolling update mechanism based on the epidemical data provided by Johns Hopkins University. First, as traditional epidemical models use the accumulative confirmed cases for training, it can only predict a rising trend of the epidemic and cannot predict when the epidemic will decline or end, an improved model is built based on long short-term memory (LSTM) with daily confirmed cases training set. Second, considering the existing forecasting model based on LSTM can only predict the epidemic trend within the next 30 days accurately, the rolling update mechanism is embedded with LSTM for long-term projections. Third, by introducing Diffusion Index (DI), the effectiveness of preventive measures like social isolation and lockdown on the spread of COVID-19 is analyzed in our novel research. The trends of the epidemic in 150 days ahead are modeled for Russia, Peru and Iran, three countries on different continents. Under our estimation, the current epidemic in Peru is predicted to continue until November 2020. The number of positive cases per day in Iran is expected to fall below 1000 by mid-November, with a gradual downward trend expected after several smaller peaks from July to September, while there will still be more than 2000 increase by early December in Russia. Moreover, our study highlights the importance of preventive measures which have been taken by the government, which shows that the strict controlment can significantly reduce the spread of COVID-19. [ABSTRACT FROM AUTHOR]

• Developed deep learning methods to forecast the COVID19 spread. • Five deep learning models have been compared for COVID-19 forecasting. • Time-series COVID19 data from Italy, Spain, France, China, the USA, and Australia are used. • Results demonstrate the potential of deep learning models to forecast COVID19 data. • Results show the superior performance of the Variational AutoEncoder model. The novel coronavirus (COVID-19) has significantly spread over the world and comes up with new challenges to the research community. Although governments imposing numerous containment and social distancing measures, the need for the healthcare systems has dramatically increased and the effective management of infected patients becomes a challenging problem for hospitals. Thus, accurate short-term forecasting of the number of new contaminated and recovered cases is crucial for optimizing the available resources and arresting or slowing down the progression of such diseases. Recently, deep learning models demonstrated important improvements when handling time-series data in different applications. This paper presents a comparative study of five deep learning methods to forecast the number of new cases and recovered cases. Specifically, simple Recurrent Neural Network (RNN), Long short-term memory (LSTM), Bidirectional LSTM (BiLSTM), Gated recurrent units (GRUs) and Variational AutoEncoder (VAE) algorithms have been applied for global forecasting of COVID-19 cases based on a small volume of data. This study is based on daily confirmed and recovered cases collected from six countries namely Italy, Spain, France, China, USA, and Australia. Results demonstrate the promising potential of the deep learning model in forecasting COVID-19 cases and highlight the superior performance of the VAE compared to the other algorithms. [ABSTRACT FROM AUTHOR]

• Developed deep learning-based models to forecast emergency department overcrowding. • Eight deep learning models have been compared for ED visits forecasting. • Real Datasets from the ED at CHRU Lille, France, investigated to serve the case study. • Two types of forecasting are conducted: one- and multi-step-ahead forecasting. • Results show the superior performance of the VAE compared to the other algorithms. As the demand for medical cares has considerably expanded, the issue of managing patient flow in hospitals and especially in emergency departments (EDs) is certainly a key issue to be carefully mitigated. This can lead to overcrowding and the degradation of the quality of the provided medical services. Thus, the accurate modeling and forecasting of ED visits are critical for efficiently managing the overcrowding problems and enable appropriate optimization of the available resources. This paper proposed an effective method to forecast daily and hourly visits at an ED using Variational AutoEncoder (VAE) algorithm. Indeed, the VAE model as a deep learning-based model has gained special attention in features extraction and modeling due to its distribution-free assumptions and superior nonlinear approximation. Two types of forecasting were conducted: one- and multi-step-ahead forecasting. To the best of our knowledge, this is the first time that the VAE is investigated to improve forecasting of patient arrivals time-series data. Data sets from the pediatric emergency department at Lille regional hospital center, France, are employed to evaluate the forecasting performance of the introduced method. The VAE model was evaluated and compared with seven methods namely Recurrent Neural Network (RNN), Long short-term memory (LSTM), Bidirectional LSTM (BiLSTM), Convolutional LSTM Network (ConvLSTM), restricted Boltzmann machine (RBM), Gated recurrent units (GRUs), and convolutional neural network (CNN). The results clearly show the promising performance of these deep learning models in forecasting ED visits and emphasize the better performance of the VAE in comparison to the other models. [ABSTRACT FROM AUTHOR]

In this paper, Deep Learning-based models are used for predicting the number of novel coronavirus (COVID-19) positive reported cases for 32 states and union territories of India. Recurrent neural network (RNN) based long-short term memory (LSTM) variants such as Deep LSTM, Convolutional LSTM and Bi-directional LSTM are applied on Indian dataset to predict the number of positive cases. LSTM model with minimum error is chosen for predicting daily and weekly cases. It is observed that the proposed method yields high accuracy for short term prediction with error less than 3% for daily predictions and less than 8% for weekly predictions. Indian states are categorised into different zones based on the spread of positive cases and daily growth rate for easy identification of novel coronavirus hot-spots. Preventive measures to reduce the spread in respective zones are also suggested. A website is created where the state-wise predictions are updated using the proposed model for authorities,researchers and planners. This study can be applied by other countries for predicting COVID-19 cases at the state or national level. [ABSTRACT FROM AUTHOR]