Application Of Long Short-Term Memory Research Articles

Application of Long Short-Term Memory (LSTM) Network for seasonal prediction of monthly rainfall across Vietnam

Application of Long Short-Term Memory (LSTM) Network for seasonal prediction of monthly rainfall across Vietnam

Composing jazz music pieces using LSTM neural networks approach

The domain of artificial intelligence has increasingly extended into the creative arts, aiming to emulate and augment human creativity with automated processes. This is particularly evident in the field of music, where AI's ability to learn and produce intricate compositions has attracted significant attention. This study explores the challenge of generating jazz music using artificial intelligence, specifically focusing on the application of Long Short-Term Memory (LSTM) neural networks for jazz composition. By optimizing the model to address the genre's complexity, the research demonstrates the LSTM's capacity to capture and reproduce jazz's essential harmonic progressions and rhythmic nuances. Quantitative analyses show high accuracy and a deep understanding of musical structures, whereas qualitative feedback confirms the model's efficacy in producing compositions that embody jazz's spontaneity. Despite its achievements, the model's tendency to generate repetitive sequences suggests areas for improvement. This paper advances the field of AI in music, illustrating the potential of LSTM networks to mimic complex musical genres and emphasizing the necessity of ongoing model refinement to foster creativity. It highlights the evolving role of machine learning in music generation, proposing a foundation for future work aimed at diminishing the gap between AI capabilities and artistic expression.

Analysis of LSTM and Derivative Models for Bitcoin Prediction Research

Bitcoin, the first cryptocurrency, has attracted much attention in the digital currency market, and its price volatility is affected by a variety of factors, posing a challenge to investors. For the sake of analytical feasibility, many studies have adopted simplified modeling assumptions. This may overlook certain key nonlinear features of the market, such as the complexity of investor behavior and the volatility of market sentiment. This study explores the application of Long Short-Term Memory (LSTM) networks and their derivative models in predicting Bitcoin prices. Recognizing the complex nature of Bitcoin’s market dynamics, the research delves into the effectiveness of LSTM in capturing the nonlinear patterns of cryptocurrency prices. Furthermore, it extends the analysis to derivative models like MSM-LSTM and Empirical Mode Decomposition (EMD) LSTM, evaluating their ability to enhance prediction accuracy by addressing the limitations of the standard LSTM. This study provides an in-depth analysis of the effectiveness of LSTM and its derived models in the practical application of Bitcoin price prediction, with a special focus on their ability to capture non-linear patterns in the market. In Bitcoin price prediction, LSTM models have limitations but also great potential in capturing nonlinear patterns in the market. Extended models based on their own can perform well in Bitcoin price prediction. This study contributes to the field by suggesting strategies to improve the accuracy of the model and by providing ideas for developing trading strategies based on the results of the analysis.

Evaluation of streamflow predictions from LSTM models in water- and energy-limited regions in the United States

The application of Long Short-Term Memory (LSTM) models for streamflow predictions has been an area of rapid development, supported by advancements in computing technology, increasing availability of spatiotemporal data, and availability of historical data that allows for training data-driven LSTM models. Several studies have focused on improving the performance of LSTM models; however, few studies have assessed the applicability of these LSTM models across different hydroclimate regions. This study investigated the single-basin trained local (one model for each basin), multi-basin trained regional (one model for one region), and grand (one model for several regions) models for predicting daily streamflow in water-limited Great Basin (18 basins) and energy-limited New England (27 basins) regions in the United States using the CAMELS (Catchment Attributes and Meteorology for Large-sample Studies) data set. The results show a general pattern of higher accuracy in daily streamflow predictions from the regional model when compared to local or grand models for most basins in the New England region. For the Great Basin region, local models provided smaller errors for most basins and substantially lower for those basins with relatively larger errors from the regional and grand models. The evaluation of one-layer and three-layer LSTM network architectures trained with 1-day lag information indicates that the addition of model complexity by increasing the number of layers may not necessarily increase the model skill for improving streamflow predictions. Findings from our study highlight the strengths and limitations of LSTM models across contrasting hydroclimate regions in the United States, which could be useful for local and regional scale decisions using standalone or potential integration of data-driven LSTM models with physics-based hydrological models.

Implementasi Algoritma Long Short-Term Memory dalam Prediksi Konsentrasi Gas Metana (CH4) di Kota Salatiga

Implementation of the Long Short Term Memory (LSTM) algorithm is done to build a prediction model that can handle complex time series data. Model development uses training and testing data and combines multiple time series to improve prediction accuracy. Model testing is done by measuring the root mean square error (RMSE) value as a performance indicator. The test results show that the application of the LSTM algorithm to the (CH4) sensor provides an optimal RMSE value, namely with a value for training data of 20% (0.09) and test data of 80% (0.14), indicating the prediction accuracy of methane gas (CH4) concentration is potentially unexploded, the results obtained have important implications for safety monitoring. This test contributes to the development of predictive methods to monitor and manage potential risks associated with (CH4) concentrations. The application of LSTM to (CH4) sensors not only improves prediction accuracy but also opens up opportunities for the development of safety systems that can more effectively predict and prevent potentially harmful phenomena due to methane gas.

Supporting Learning Information System through Knowledge Management Optimization using Long Short-Term Memory Method

Effective information and knowledge management is vital in many areas, including higher education. The use of artificial intelligence (AI) technology, especially the long short-term memory (LSTM) information system performance patterns in the educational world. This article explores the application of LSTM to optimize knowledge management in colleges, focusing on the prediction of information systems performance. The proposed methods include text classification steps, with measures such as data collection, data pre-processing, word representation, classification, and evaluation. The test results showed that the LSTM model managed to classify reviews labeled positive, neutral, and negative with an accuracy of 33.33%. However, the success of the model was limited by the size of the data set and the pre-processing involved. This research recommends further development with the addition of experimental data, proper preprocessing adjustments, and better hyperparameter identification to improve the accuracy of the prediction results.
 Keywords: information management, artificial intelegence, LSTM, text classification, knowledge management, accurate prediction

Empowering Microgrid Energy Management with Artificial Intelligence and Model Predictive Control

This paper presents an innovative approach to microgrid energy management by integrating Model Predictive Control (MPC) with Artificial Intelligence (AI), focusing on the application of Long Short-Term Memory (LSTM) networks for load forecasting. We show that AI-enhanced MPC can significantly improve the efficiency and reliability of microgrid energy management. The fundamental results of the LSTM models highlight the effectiveness of our methodology in improving predictive accuracy and operational performance.

Deep learning-based forecasting modeling of micro gas turbine performance projection: An experimental approach

Performance forecasting of aeroengines is crucial for achieving better operational efficiency, ensuring safety, reducing costs, and minimizing environmental impact in the aviation industry. It enables engineers and researchers to make informed decisions, leading to advancements in technology and the overall evolution of aviation. This study is focused on the effects of flight conditions on the performance of a turbojet, which in consequence affects the environmental aspect of operation. By investigating the relationships between aeroengine efficiency and performance indicators such as thrust, shaft speed, and exhaust gas temperature (EGT), and flight characteristics expressed in terms of environmental and operational conditions, the study seeks to elucidate these connections. The article's significance lies in its successful application of Long Short-Term Memory (LSTM) networks to predict thrust, shaft speed, and EGT variations in turbojet engines under varying flight conditions. Experimental data from a turbojet test bench is processed with deep learning, specifically LSTM recurrent neural networks that are developed based on Matrix Laboratory (MATLAB). The model inputs are free stream air speed, compressor inlet pressure, combustor inlet temperature, combustor inlet pressure, turbine inlet temperature, turbine inlet pressure, nozzle inlet pressure and fuel flow, and the outputs are thrust, shaft speed and EGT. Predicted thrust closely aligns with actual thrust values, though with minor discrepancies. Shaft speed predictions exhibit a similar trend, while EGT predictions showcase a comparable pattern with slight variations. Despite the prediction errors, a thorough evaluation of median values, box plots, and probability density functions confirms that the models effectively capture available information, though discrepancies may arise from measurement inaccuracies and initial engine conditions. These results show that it is possible to accurately predict turbojet performance using LSTM recurrent neural network. This research paves the way for enhanced aeroengine performance prediction, particularly in scenarios requiring off-board or high-performance applications.

Stock Market Analysis and Prediction Using LSTM: A Case Study on Technology Stocks

This research explores the application of Long Short-Term Memory (LSTM) networks for stock market analysis and prediction, focusing on four major technology stocks: Apple Inc. (AAPL), Google LLC (GOOG), Microsoft Corporation (MSFT), and Amazon.com Inc. (AMZN). Historical stock price data from Yahoo Finance spanning from January 1, 2012, to the present is utilized. The study aims to develop and evaluate an LSTM-based prediction model for forecasting future stock prices. The LSTM model consists of two LSTM layers with 128 and 64 units, respectively, followed by two dense layers. The model is trained using the Adam optimizer and mean squared error (MSE) loss function. Evaluation of the model is done using the root mean squared error (RMSE) metric. The results demonstrate the potential of LSTM models in capturing complex patterns in stock price movements and making reasonably accurate predictions.

Machine Learning-Enhanced Decision Support for Neurological Disorders Management in Healthcare

In a nutshell: Neurological illnesses, which include a wide range of ailments from Parkinson's disease to epilepsy, present considerable problems to the administration of healthcare. The purpose of this abstract is to present an overview of a unique technique called Machine Learning-Enhanced Decision Support (ML-EDS) for Neurological Disorders Management, which aims to improve patient treatment as well as patient outcomes. Within the scope of this investigation, we have suggested a complete ML-EDS framework that integrates cutting-edge machine learning strategies with more conventional approaches to medical treatment. The framework provides a comprehensive solution that makes use of several types of patient data, such as clinical records, neuroimaging data, genetic information, and data from wearable sensors. The Random Forest, Long Short-Term Memory (LSTM), and k-Nearest Neighbours (k-NN) algorithms are integrated into our methodology in order to improve illness progression prediction, seizure event detection, and individualized medication suggestion. In comparison to more conventional approaches, the suggested framework reveals noteworthy advantages. When compared to conventional algorithms, it demonstrates superior accuracy, sensitivity, and specificity in the prediction of illness development. Additionally, its F1 score is significantly higher. In addition, the application of LSTM for the identification of seizure episodes demonstrates the effectiveness of the method in determining crucial occurrences, which in turn enables prompt intervention. The possibility of tailored medical treatment based on k-means clustering is also highlighted in this research. Patients with comparable clinical characteristics can be grouped together to facilitate the development of personalized treatment methods, which can then provide individualized care pathways. The performance evaluation demonstrates that k-NN is capable of making therapy recommendations with high precision, hence contributing to an overall improvement in the health of the patient. Additionally, the proposed framework performs very well in terms of computational efficiency, reducing the amount of time required for both training and prediction, which positions it as a viable option for healthcare applications that are used in the real world. In summation, the use of machine learning in decision support for the management of neurological disorders in healthcare is a ground-breaking strategy that is reshaping the therapeutic landscape. It does this by using the power of machine learning, which enables it to provide more accurate illness progression predictions, rapid seizure event detection, and highly tailored medical care. The findings and efficiency metrics conclusively indicate that this strategy is superior to existing procedures, suggesting a paradigm change in the therapy of neurological disorders.

Advancing Agricultural Crop Recognition: The Application of LSTM Networks and Spatial Generalization in Satellite Data Analysis

This study addresses the challenge of accurate crop detection using satellite data, focusing on the application of Long Short-Term Memory (LSTM) networks. The research employs a “spatial generalization” approach, where a model trained on one geographic area is applied to another area with similar vegetation conditions during the growing season. LSTM networks, which are capable of learning long-term temporal dependencies, are used to overcome the limitations of traditional machine learning techniques. The results indicate that LSTM networks, although more computationally expensive, provide a more accurate solution for crop recognition compared with other methods such as Multilayer Perceptron (MLP) and Random Forest algorithms. The accuracy of LSTM networks was found to be 93.7%, which is significantly higher than the other methods. Furthermore, the study showed a high correlation between the real and model areas of arable land occupied by different crops in the municipalities of the study area. The main conclusion of this research is that LSTM networks, combined with a spatial generalization approach, hold great promise for future agricultural applications, providing a more efficient and accurate tool for crop recognition, even in the face of limited training data and complex environmental variables.

Stock Price Forecasting with LSTM: A Brief Analysis of Mathematics Behind LSTM

Stock marketing prediction is really important in this digital world and for this reason, its forecasting tools are highly desired. This research examines the influence of Long Short-Term Memory (LSTM) in stock market prediction. The basic genetic algorithm behind the time series analysis tool is described with the help of a structure diagram, and its mathematical structure is explained. We also described the application of LSTM in the field of finance. LSTM is playing a vital role in prediction in several applications. Experiments on the proposed method have been conducted on stock market data. The results show that LSTM predicts the price quite fairly accurately. It outperformed other prediction models due to long-term dependency and due to its accuracy. The research demonstrates the LSTM tool’s impact on finance, particularly stock market prediction.

Long short-term memory models to quantify long-term evolution of streamflow discharge and groundwater depth in Alabama

Long short-term memory (LSTM) models have been shown to be efficient for rainfall-runoff modeling, and to a lesser extent, for groundwater depth forecasting. In this study, LSTMs were applied to quantify the spatiotemporal evolution of surface and subsurface hydrographs in Alabama in the Southeastern United States, where water sustainability has not been fully quantified across spatiotemporal scales. First, the surface water LSTM model with extensive dynamic (precipitation and other weather variables) and static (basin characteristics) inputs predicted the main characteristics of streamflow for six years at 19 gauged basins in Alabama. The model tended to underestimate extremely high streamflow but adding drainage density as an input feature slightly improved the predictions of extreme events. Second, to predict the groundwater depth evolution, a groundwater LSTM (GW-LSTM) model was proposed and applied using static inputs capturing the aquifers' hydrogeological properties and dynamic inputs of meteorological information. Three precipitation scenarios were also explored to evaluate the groundwater hydrograph evolution in the next two decades. The GW-LSTM model predicted the general trend of daily groundwater depth fluctuations (at 21 wells distributed across Alabama from 1990 to 2021) including most extremely high groundwater levels, and recovered groundwater depth for locations withheld from model training and validation. This study, therefore, extended the application of LSTMs in quantifying the spatiotemporal evolution of surface water and groundwater, two manifestations of a single integrated resource.

Deep-learning-based data loss reconstruction for spatiotemporal temperature in piloti structures: Enhancing applicability with limited datasets

A time-temperature curve, representing the fire characteristics of structures, can be obtained by real fire experiments. However, these experiments are inherently susceptible to data loss, which can compromise the accuracy of results. To address this challenge, this study proposed the framework utilizing a long-short-term memory (LSTM) with Bayesian optimization to reconstruct temperature histories by learning the spatiotemporal correlation of the data. The proposed framework is first validated using simulated datasets from computational fluid dynamics analyses. The field applicability of the model is further demonstrated through real fire test results, affirming its reliability in practical scenarios. The study also introduces a novel data processing technique to mitigate overfitting issues in LSTM applications with limited data, enhancing the robustness and reliability of temperature history reconstruction. Overall, the results highlight the potential of deep learning in accurately and practically reconstructing temperature histories in fire experiments.

An International Study of Application of Long Short-Term Memory (LSTM) Neural Networks for the prediction of stock and forex markets

This study focuses on the use of Long Short-Term Memory (LSTM) neural networks for stock and currency market forecasting. Accurate projections are difficult to make because of the complex dynamics and non-linear interactions that characterise financial markets. A Recurrent Neural Networks (RNN) variation called LSTM is particularly good at identifying temporal dependencies and long-term patterns in sequential data. The goal of LSTM models is to discover significant insights and produce precise predictions using past price data. In addition to discussing data pretreatment methods, model creation, and assessment metrics related to stock and FX market prediction, this work examines the benefits of LSTM in capturing market dynamics. Case studies and empirical analysis are used to investigate the capabilities and constraints of LSTM models in forecasting market movements.

Daily Peak-Electricity-Demand Forecasting Based on Residual Long Short-Term Network

Forecasting the electricity demand of buildings is a key step in preventing a high concentration of electricity demand and optimizing the operation of national power systems. Recently, the overall performance of electricity-demand forecasting has been improved through the application of long short-term memory (LSTM) networks, which are well-suited to processing time-series data. However, previous studies have focused on improving the accuracy in forecasting only overall electricity demand, but not peak demand. Therefore, this study proposes adding residual learning to the LSTM approach to improve the forecast accuracy of both peak and total electricity demand. Using a residual block, the residual LSTM proposed in this study can map the residual function, which is the difference between the hypothesis and the observed value, and subsequently learn a pattern for the residual load. The proposed model delivered root mean square errors (RMSE) of 10.5 and 6.91 for the peak and next-day electricity demand forecasts, respectively, outperforming the benchmark models evaluated. In conclusion, the proposed model provides highly accurate forecasting information, which can help consumers achieve an even distribution of load concentration and countries achieve the stable operation of the national power system.

How can we benefit from regime information to make more effective use of long short-term memory (LSTM) runoff models?

Abstract. To date, long short-term memory (LSTM) networks have been successfully applied to a key problem in hydrology: the prediction of runoff. Unlike traditional conceptual models, LSTM models are built on concepts that avoid the need for our knowledge of hydrology to be formally encoded into the model. The question, then, is how we can still make use of our domain knowledge and traditional practices, not to build the LSTM models themselves, as we do for conceptual models, but to use them more effectively. In the present paper, we adopt this approach, investigating how we can use information concerning the hydrologic characteristics of catchments for LSTM runoff models. In this first application of LSTM in a French context, we use 361 gauged catchments with very diverse hydrologic conditions from across France. The catchments have long time series of at least 30 years. Our main directions for investigation include (a) the relationship between LSTM performance and the length of the LSTM input sequence within different hydrologic regimes, (b) the importance of the hydrologic homogeneity of catchments when training LSTMs on a group of catchments, and (c) the interconnected influence of the local tuning of the two important LSTM hyperparameters, namely the length of the input sequence and the hidden unit size, on the performance of group-trained LSTMs. We present a classification built on three indices taken from the runoff, precipitation, and temperature regimes. We use this classification as our measure of homogeneity: catchments within the same regime are assumed to be hydrologically homogeneous. We train LSTMs on individual catchments (local-level training), on catchments within the same regime (regime-level training), and on the entire sample (national-level training). We benchmark local LSTMs using the GR4J conceptual model, which is able to represent the water gains/losses in a catchment. We show that LSTM performance has the highest sensitivity to the length of the input sequence in the Uniform and Nival regimes, where the dominant hydrologic process of the regime has clear long-term dynamics; thus, long input sequences should be chosen in these cases. In other regimes, this level of sensitivity is not found. Moreover, in some regimes, almost no sensitivity is observed. Therefore, the size of the input sequence in these regimes does not need to be large. Overall, our homogeneous regime-level training slightly outperforms our heterogeneous national-level training. This shows that the same level of data adequacy with respect to the complexity of representation(s) to be learned is achieved in both levels of training. We do not, however, exclude a potential role of the regime-informed property of our national LSTMs, which use previous classification variables as static attributes. Last but not least, we demonstrate that the local selection of the two important LSTM hyperparameters (the length of the input sequence and the hidden unit size) combined with national-level training can lead to the best runoff prediction performance.

Application of LSTM based on the BAT-MCS for binary-state network approximated time-dependent reliability problems

Reliability is an important tool for evaluating the performance of modern networks. Currently, it is NP-hard and #P-hard to calculate the exact reliability of a binary-state network when the reliability of each component is assumed to be fixed. However, this assumption is unrealistic because the reliability of each component always varies with time. To meet this practical requirement, we propose a new algorithm called the binary-addition-tree algorithm and Monte Carlo simulation based Long Short-Term Memory (LSTM-BAT-MCS), based on long short-term memory (LSTM), the Monte Carlo simulation (MCS), and the binary-addition-tree algorithm (BAT). The superiority of the proposed LSTM-BAT-MCS was demonstrated by experimental results of three benchmark networks with at most 10−4 mean square error.

Utilization of the Long Short-Term Memory network for predicting streamflow in ungauged basins in Korea

Hydrological prediction is essential for managing and preserving headwater, wetland, and rural basins, yet it is difficult due to a lack of data. The Long Short-Term Memory (LSTM) network is a promising deep learning approach and has demonstrated excellent performance in streamflow prediction. However, due to the characteristics that require abundant and high-quality observations, most LSTM applications focus on investigating performance for data-rich basins. To investigate the applicability of LSTM in ungauged basins without hydrological observations, we introduce and test an approach to predict streamflow in ungauged basins using the LSTM network that has learned the integration of data from multiple gauged basins. Four learning strategies are constructed using various datasets such as ground-observed meteorological data, satellite data, and hydro-geomorphological characteristics. As a result, the LSTM network that learned the meteorological data in multiple gauged basins satisfactorily predicted streamflow in ungauged basins. The LSTM network that learned the satellite data further showed affirmative results. On the other hand, the LSTM, which additionally learned the hydro-geomorphological characteristics of the basins, exposed the need for improvement, such as securing additional various types of training data. In addition, it was recognized that additional efforts were needed to solve overfitting and out-of-distribution prediction problems. However, this approach achieved model performance deriving metrics above threshold values. These results show the potential of LSTM for streamflow prediction in ungauged basins including wetlands.

Application of Long Short-Term Memory (LSTM) on the Prediction of Rainfall-Runoff in Karst Area

In actual engineering fields, the bearing capacity of a rock is closely related to the pore water pressure in the rock. Studies have shown that the pore water in the rock has a great relationship with the change in runoff. Thus, it has crucial meaning to accurately evaluate and quantitate the property of the rainfall–runoff, and many traditional classic models are proposed to study the characteristic of rainfall–runoff. While considering the high uncertainty and randomness of the rainfall–runoff property, more and more artificial neural networks (ANN) are used for the rainfall–runoff modeling as well as other fields. Among them, the long short-term memory (LSTM), which can be trained for sequence generation by processing real data sequences one step at a time and has good prediction results in other engineering fields, is adopted in this study to investigate the changes of rainfall–runoff values and make a prediction. In order to ensure the accuracy of the trained model, the cross-validation method is used in this study. The training data set is divided into 12 parts. The monthly forecast results from 2014 to 2015 show that the model can well reflect the peaks and troughs. In a recent study, the relationship between the rainfall–runoff and discharge are commonly based on the current measured data, while the prediction results are adopted to analyze the relation of these parameters, and considering that the existing methods have fuzzy relationship between runoff and discharge, which leads to a high risk of forecasting and dispatching. A method of modeling analysis and parameter estimation of hydrological runoff and discharge relationship based on machine learning is designed. From the experimental results, the average risk of this method is 61.23%, which is 15.104% and 13.397% less than that of the other two existing methods, respectively. It proves that the method of hydrological runoff relationship modeling and parameter estimation integrated with machine learning has better practical application effect.