Deep recurrent models for forecasting infectious diseases

In recent years, the penetration of solar power at residential and utility levels has progressed exponentially. However, due to its stochastic nature, the prediction of solar global horizontal irradiance (GHI) with higher accuracy is a challenging task; but, vital for grid management: planning, scheduling & balancing. Therefore, this paper proposes an ensemble model using the extended scope of wavelet transform (WT) and bidirectional long short term memory (BiLSTM) deep learning network to forecast 24-h ahead solar GHI. The WT decomposes the input time series data into different finite intrinsic model functions (IMF) to extract the statistical features of input time series. Further, the study reduces the number of IMF series by combining the wavelet decomposed components (D1-D6) series on the basis of comprehensive experimental analysis with an aim to improve the forecasting accuracy. Next, the trained standalone BiLSTM networks are allocated to each IMF sub-series to execute the forecasting. Finally, the forecasted values of each sub-series from BiLSTM networks are reconstructed to deliver the final solar GHI forecast. The study performed monthly solar GHI forecasting for one year dataset using one month moving window mechanism for the location of Ahmedabad, Gujarat, India. For the performance comparison, the naïve predictor as a benchmark model, standalone long short term memory (LSTM), gated recurrent unit (GRU), BiLSTM and two other wavelet-based BiLSTM models are also simulated. From the results, it is observed that the proposed model outperforms other models in terms of root mean square error (RMSE) & mean absolute percentage error (MAPE), coefficient of determination (R2) and forecast skill (FS). The proposed model reduces the monthly average RMSE by range from 26.04–58.89%, 5.17–31.35%, 23.26–56.06% & 21.08–57% in comparison with benchmark, standalone BiLSTM, GRU & LSTM networks respectively. On the other hand, the monthly average MAPE is reduced by range from 9 to 51.18%, 12.59–28.14%, 30.43–59.19% & 26.54–58.92% in comparison to benchmark, standalone BiLSTM, GRU & LSTM respectively. Further, the proposed model obtained the value of R2 equal to 0.94 and forecast skill (%) of 47% with reference to the benchmark model.

Time Series Forecasting of Environmental Dynamics in urban forests is quite challenging, unless new approaches such as deep learning and remote sensing are employed. Deep learning-based time series algorithms offer robust scientific capabilities for forecasting and assessing sustainability trends using sequential data. Among these, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Bidirectional LSTM (BiLSTM) have gained widespread adoption across various predictive modeling domains. In the present research, these algorithms are employed to analyze urban forest raster data derived from the Srengseng Ecotourism Forest, located in West Jakarta, Indonesia. The present study focuses on predicting the temporal patterns of key spatial indicators: Normalized Difference Vegetation Index (NDVI), Land Surface Temperature (LST), and Forest Cover Density (FCD) in the Srengseng urban ecotourism forest area, spanning the years 2014 to 2024, through the application of LSTM, GRU, and BiLSTM deep learning architectures. The methodology used in this study is a combined approach involving remote sensing and deep learning. Spatial data were acquired through the delineation of a high-precision polygon of Srengseng Urban Forest using Google Earth Pro and Google Earth Engine (GEE). GeoTIFF datasets of NDVI, LST, and FCD for the years 2014–2024 were processed using Python-based modeling scripts. Model performance was evaluated through a comparative analysis of LSTM, GRU, and BiLSTM in predicting temporal trends in these ecological indicators. The results of this study show that the Bidirectional LSTM (BiLSTM) consistently demonstrated superior performance to predict all the data spatially, with scores of 0.94 for NDVI, 0.90 for FCD, and 0.85 for LST. Followed by LSTM that predicts NDVI (0.87), FCD (0.89), LST (0.83), as well as GRU, which can estimate spatial data NDVI (0.86), FCD (0.89), and LST (0.85). These results outperformed the predictive accuracy of both the standard LSTM and GRU models.

This study explores the effectiveness of various deep learning models for detecting spam in YouTube comments. Six models were evaluated: Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), Gated Recurrent Unit (GRU), and Attention mechanisms. The dataset consists of 1,956 real comments extracted from popular YouTube videos, representing both spam and legitimate messages. The preprocessing phase involved tokenization and padding of text sequences to prepare them for model input. Results reveal that the LSTM model achieved the highest test accuracy of 95.65%, outperforming other models by capturing sequential dependencies and context within comments. The CNN model also demonstrated high accuracy, underscoring the importance of local pattern recognition in text classification. While BiLSTM and Attention models offered comparable performance, their marginal improvement over LSTM indicates that sequential modeling plays a crucial role in this task. The GRU model, despite being computationally efficient, showed slightly lower accuracy compared to LSTM and BiLSTM. The MLP model, serving as a baseline, exhibited limited performance, emphasizing the need for advanced architectures in spam detection. These findings suggest that combining sequential modeling with local feature extraction could lead to more robust spam detection systems.

Stratification of Depressed and Non-Depressed Texts from Social Media using LSTM and its Variants

Highly accurate cryptocurrency price predictions are of paramount interest to investors and researchers. However, owing to the nonlinearity of the cryptocurrency market, it is difficult to assess the distinct nature of time-series data, resulting in challenges in generating appropriate price predictions. Numerous studies have been conducted on cryptocurrency price prediction using different Deep Learning (DL) based algorithms. This study proposes three types of Recurrent Neural Networks (RNNs): namely, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Bi-Directional LSTM (Bi-LSTM) for exchange rate predictions of three major cryptocurrencies in the world, as measured by their market capitalization—Bitcoin (BTC), Ethereum (ETH), and Litecoin (LTC). The experimental results on the three major cryptocurrencies using both Root Mean Squared Error (RMSE) and the Mean Absolute Percentage Error (MAPE) show that the Bi-LSTM performed better in prediction than LSTM and GRU. Therefore, it can be considered the best algorithm. Bi-LSTM presented the most accurate prediction compared to GRU and LSTM, with MAPE values of 0.036, 0.041, and 0.124 for BTC, LTC, and ETH, respectively. The paper suggests that the prediction models presented in it are accurate in predicting cryptocurrency prices and can be beneficial for investors and traders. Additionally, future research should focus on exploring other factors that may influence cryptocurrency prices, such as social media and trading volumes.

In this paper, we have implemented Stacked Recurrent Neural Network for a Robust Speech Recognition System inside a car. To cancel out the high traffic noise, Long Short Term Memory (LSTM) and Gated Recurrent Unit (GRU), Bidirectional LSTM (BLSTM) and Bidirectional GRU (BGRU) have been used for mapping the noisy data with the clean speech signal. Later to increase the complexity of the model, Stacked LSTM, GRU, BLSTM and BGRU were implemented. All the models were applied to the data in Cepstral Domain and analyzed with respect to different Number of Layers and different Signal to Noise Ratio (SNR).

The work presented in this paper aims to improve the accuracy of forecasting models in univariate time series, for this it is experimented with different hybrid models of two and four layers based on recurrent neural networks such as Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU). It is experimented with two time series corresponding to downward thermal infrared and all sky insolation incident on a horizontal surface obtained from NASA’s repository. In the first time series, the results achieved by the two-layer hybrid models (LSTM + GRU and GRU + LSTM) outperformed the results achieved by the non-hybrid models (LSTM + LSTM and GRU + GRU); while only two of six four-layer hybrid models (GRU + LSTM + GRU + LSTM and LSTM + LSTM + GRU + GRU) outperformed non-hybrid models (LSTM + LSTM + LSTM + LSTM and GRU + GRU + GRU + GRU). In the second time series, only one model (LSTM + GRU) of two hybrid models outperformed the two non-hybrid models (LSTM + LSTM and GRU + GRU); while the four-layer hybrid models, none could exceed the results of the non-hybrid models.

An accurate and efficient forecasting of solar energy is necessary for managing the electricity generation and distribution in today’s electricity supply system. However, due to its random character in its time series, accurate forecasting of solar irradiation is a difficult task; but it is important for grid management, scheduling and its balancing. To fully utilize the solar energy in order to balance the generation and consumption, this paper proposed an ensemble approach using CEEMDAN-BiLSTM combination to forecast short term solar irradiation. In this, Complete Ensemble Empirical Mode Decomposition with adaptive noise (CEEMDAN) extract the inherent characteristics of time series data by decomposing it into low and high frequency Intrinsic Mode Functions (IMF’s) and Bidirectional Long Short Term Memory (BiLSTM) used as a forecasting tool to forecast the solar Global Horizontal Irradiance (GHI). Furthermore, using extensive experimental analysis, the research minimizes the number of IMF’s by integrating the CEEMDAN decomposed component (IMF1–IMF14) in order to increase the prediction accuracy. Then, for each IMF subseries, the trained standalone BiLSTM network are assigned to carry out the forecasting. In last stage, the forecasted results of each BiLSTM network are aggregate to compile final results. Two year data (2012–13) of Delhi, India from National Solar Radiation Database (NSRDB) has been used for training while one year data (2014) used for testing purpose for the same location. The proposed model performance is measured in terms of root mean square error (RMSE), mean absolute percentage error (MAPE), Correlation coefficient (R22) and forecast skill (FS). For the comparative analysis of proposed model, several others models: persistence model, unidirectional deep learning models: long short term memory (LSTM), gated recurrent unit (GRU), BiLSTM and two CEEMDAN based BiLSTM models are developed. The proposed model achieved lowest annual average RMSE (18.86 W/m22, 22.24 W/m22, 26.25 W/m22) and MAPE (2.19%, 4.81%, 6.77%) among the other developed models for 1-hr, 2-hr and 3-hr ahead solar GHI forecasting respectively. The maximum correlation coefficient (R22) obtained by the proposed model is 96.4 for 1-hr ahead respectively; on the other hand, forecast skill (%) of 89% with reference to benchmark model. Various test such as: Diebold Mariano Hypothesis test (DMH) and directional change in forecasting (DC) are used to analyze the sensitivity with reference to the difference in forecasted and observed value.

As a new type of currency introduced in the new millennium, cryptocurrency has established its ecosystems and attracts many people to use and invest in it. However, cryptocurrencies are highly dynamic and volatile, making it challenging to predict their future values. In this research, we use a multivariate prediction approach and three different recurrent neural networks (RNNs), namely the long short-term memory (LSTM), the bidirectional LSTM (Bi-LSTM), and the gated recurrent unit (GRU). We also propose simple three layers deep networks architecture for the regression task in this study. From the experimental results on five major cryptocurrencies, i.e., Bitcoin (BTC), Ethereum (ETH), Cardano (ADA), Tether (USDT), and Binance Coin (BNB), we find that both Bi-LSTM and GRU have similar performance results in terms of accuracy. However, in terms of the execution time, both LSTM and GRU have similar results, where GRU is slightly better and has lower variation results on average.

In the context of global climate change and the continuous development of urban areas, rainfall-inundation modeling is a common approach that provides critical support for the protection and early warning of urban waterlogging protection. The present study conducts a data-driven model for hourly urban rainfall-inundation depth prediction, which is based on a gated recurrent unit (GRU) neural network and uses the simulated annealing (SA) algorithm for the hyperparameter optimization of GRU, namely the SA-GRU model. To verify the performance of the proposed model, backpropagation, long short-term memory (LSTM), and bidirectional LSTM (BiLSTM) neural networks are set as benchmarks. Results show that the SA-GRU has high accuracy in the case of short-term inundation prediction, with the Nash–Sutcliffe efficiency from 0.999 to 0.596 for the 1-h-ahead to 8-h-ahead predictions. And further research reveals that the SA-GRU integrates the significant optimization of SA, with an average 20% reduction of the root mean square error within the first eight prediction periods, and the efficient training speed of GRU, with 23.7% faster than LSTM and 44.2% faster than BiLSTM. In conclusion, the SA-GRU excels in urban inundation prediction, demonstrating its value in flood management and decision-making.

The accurate forecast of wastewater treatment plant (WWTP) key features can comprehend and predict the plant behavior to support process design and controls, improve system reliability, reduce operational costs, and endorse optimization of overall performances. Deep learning technologies as proven data-driven soft-sensors should be developed for WWTP applications to tackle the process of non-linearity and the dynamic nature of environmental data. This study adopts deep learning-based models as soft-sensors to forecast WWTP key features, such as influent flow, influent temperature, influent biochemical oxygen demand (BOD), effluent chloride, effluent BOD, and power consumption. We constructed six deep learning models derived from long short-term memory (LSTM) and gated recurrent unit (GRU), namely traditional LSTM and GRU, the exponentially smoothed LSTM, and the adaptive version of LSTM and smoothed LSTM. The employment of a smoothed LSTM technique is expected to reduce the outlier effect and to improve forecasting accuracy. Meanwhile, the usage of adaptive deep models will enhance the capabilities of the LSTM to quickly and accurately follow the trend of future data. We compared the performance of these models with Bi-directional LSTM (BiLSTM) and the seasonal decomposition using local regression. The historical records from a coastal municipal WWTP in Saudi Arabia are used to verify the investigated models' effectiveness. The proposed models provide promising forecasting results but require no assumptions on the data distributions. In terms of efficiency, GRU based models converge faster than LSTM based models. In terms of accuracy, the LSTM soft-sensor shows overall the optimal result for all key features followed by the exponentially-smoothed GRU and LSTM. By contrast, the adaptive models achieved the lowest forecasting performance compared to the other models. These findings will benefit practitioners to achieve data-driven WWTP management.

Conventional approaches to analyzing power losses in electrical transmission networks have largely emphasized generic power loss minimization through the integration of loss-reducing devices such as shunt capacitors. However, achieving optimal power loss minimization requires a more data-driven and intelligent approach that transcends traditional methods. This study presents a novel classification-based methodology for detecting and analyzing transmission line losses using real-world data from the Ikorodu–Sagamu 132 kV double-circuit line in Nigeria, selected for its dense concentration of high-voltage consumers. Twelve (12) transmission lines were examined, and the collected data were subjected to comprehensive preprocessing, feature engineering, and modeling. The classification capabilities of advanced deep learning models—Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (BiLSTM), and Gated Recurrent Unit (GRU)—were explored through six experimental scenarios: LSTM, LSTM with Attention Mechanism (LSTM-AM), BiLSTM, GRU, LSTM-BiLSTM, and LSTM-GRU. These models were implemented using the Python programming environment and evaluated using standard performance metrics, including accuracy, precision, recall, F1-score, support, and confusion matrices. Statistical analysis revealed significant variability in transmission losses, particularly in lines such as I1, Ps, Ogy, and ED, which exhibited high standard deviations. The LSTM-AM model achieved the highest classification accuracy of 83.84%, outperforming both standalone and hybrid models. In contrast, BiLSTM yielded the lowest performance. The findings demonstrate that while standalone models like GRU and LSTM are effective, the incorporation of attention mechanisms into LSTM architecture enhances classification accuracy. This study provides a compelling case for employing deep learning-based classification techniques in intelligent power loss classification across transmission networks. It also supports the realization of SDG 7 by aiming to provide access to reliable, affordable, and sustainable energy for all.

Accurate landslide early warning systems are a trustworthy risk-reduction method that may greatly minimize human and economic losses. Several machine learning algorithms have been investigated for this goal, underlying the impressive potential in prediction capability of Deep Learning (DL) models. Despite this, the only DL models evaluated so far are the long short-term memory (LSTM) and Gated Recurrent Unit (GRU) algorithms. Several alternative DL algorithms, however, are appropriate for time series forecasting problems. In this research, we evaluate, analyze, and present seven DL approaches for the forecasting of landslide displacement: LSTM, 2xLSTM, bidirectional LSTM (Bi-LSTM),Multilayer perception (MLP), 1D convolutional neural network (1D CNN), GRU, and an architecture build of 1D CNN and LSTM (Conv-LSTM). The study examines four different landslides with varying geographical locations, geological conditions, time step size, and measuring devices. Two landslides are placed in an artificial reservoir scenario, whereas the other two are affected only by rainfall. The findings show that the MLP, GRU, and LSTM models can produce accurate predictions in all four situations, with the Conv-LSTM model outperforming the others in the Baishuihe landslide, which is extremely seasonal. There are no discernible variations in performance between landslides within and outside constructed reservoirs. Furthermore, the study finds that MLP is better suited to forecasting the largest displacement peaks, whilst LSTM and GRU are better suited to forecasting smaller displacement peaks. We feel that the outcomes of this study will be extremely beneficial in developing a DL-based landslide early warning system (LEWS).

Accurate early warning systems for landslides are a reliable risk-reduction strategy that may significantly reduce fatalities and economic losses. Several machine learning methods have been examined for this purpose, underlying deep learning (DL) models’ remarkable prediction capabilities. The long short-term memory (LSTM) and gated recurrent unit (GRU) algorithms are the sole DL model studied in the extant comparisons. However, several other DL algorithms are suitable for time series forecasting tasks. In this paper, we assess, compare, and describe seven DL methods for forecasting future landslide displacement: multi-layer perception (MLP), LSTM, GRU, 1D convolutional neural network (1D CNN), 2xLSTM, bidirectional LSTM (bi-LSTM), and an architecture composed of 1D CNN and LSTM (Conv-LSTM). The investigation focuses on four landslides with different geographic locations, geological settings, time step dimensions, and measurement instruments. Two landslides are located in an artificial reservoir context, while the displacement of the other two is influenced just by rainfall. The results reveal that the MLP, GRU, and LSTM models can make reliable predictions in all four scenarios, while the Conv-LSTM model outperforms the others in the Baishuihe landslide, where the landslide is highly seasonal. No evident performance differences were found for landslides inside artificial reservoirs rather than outside. Furthermore, the research shows that MLP is better adapted to forecast the highest displacement peaks, while LSTM and GRU are better suited to model lower displacement peaks. We believe the findings of this research will serve as a precious aid when implementing a DL-based landslide early warning system (LEWS).

Traditional research mostly focuses on the analysis of static data, such as note sequence, velocity, force, etc., while neglecting the dynamic process in piano performance. Piano performance is a dynamic art form that includes finger movements, coherence between notes, musical expression, etc., which are often overlooked in traditional research. This article uses BiLSTM (Bidirectional Long Short-Term Memory) to dynamically analyze the piano performance process, in order to better understand the dynamic characteristics of finger movement trajectory and improve the effectiveness of piano performance. Note sequence data is collected and the direction of finger movement is labeled, which includes four directions: up, down, left, and right. The BiLSTM model is trained using annotated datasets, and the note sequence is partitioned using time windows. The direction of finger movement on the keyboard is predicted using BiLSTM. The experiment compared BiLSTM with LSTM (Long Short-Term Memory) and GRU (Gated Recurrent Unit), and found that the average MSE (Mean Squared Error) for finger motion trajectory prediction of BiLSTM was 0.08 and the average RAE (Relative Absolute Error) was 0.06. The prediction accuracy of BiLSTM in the four directions of up, down, left, and right reached 98.2%, 97.4%, 97.2%, and 97.5%, respectively. Therefore, using BiLSTM can accurately predict finger movement trajectories and effectively analyze piano performance techniques.