Search Results (97)

Modelling events that change over time is one of the most difficult problems in data analysis. Forecasting of time-varying electric power values is also an important problem in data analysis. Regression methods, machine learning, and deep learning methods are used to learn different patterns from data and develop a consumption prediction model. The aim of this study is to determine the most successful models for short-term power consumption prediction with deep learning and to achieve the highest prediction accuracy. In this study, firstly, the data was evaluated and organized with exploratory data analysis (EDA) on a ready dataset and the features of the data were extracted. Studies were carried out on long short-term memory (LSTM), gated recurrent unit (GRU), simple recurrent neural networks (SimpleRNN) and bidirectional long short-term memory (BiLSTM) architectures. First, four architectures were used with 11 different optimization methods. In this study, it was seen that a high success rate of 0.9972 was achieved according to the R² score index. In the following, the first study was tried with different epoch numbers. Afterwards, this study was carried out with 264 separate models produced using four architectures, 11 optimization methods, and six activation functions in order. The results of all these studies were obtained according to the root mean square error (RMSE), mean absolute error (MAE), and R²_score indexes. The R²_score indexes graphs are presented. Finally, the 10 most successful applications are listed. Full article

This study investigates the application of advanced deep learning models for the classification of aviation safety incidents, focusing on four models: Simple Recurrent Neural Network (sRNN), Gated Recurrent Unit (GRU), Bidirectional Long Short-Term Memory (BLSTM), and DistilBERT. The models were evaluated based on key performance metrics, including accuracy, precision, recall, and F1-score. DistilBERT achieved perfect performance with an accuracy of 1.00 across all metrics, while BLSTM demonstrated the highest performance among the deep learning models, with an accuracy of 0.9896, followed by GRU (0.9893) and sRNN (0.9887). Class-wise evaluations revealed that DistilBERT excelled across all injury categories, with BLSTM outperforming the other deep learning models, particularly in detecting fatal injuries, achieving a precision of 0.8684 and an F1-score of 0.7952. The study also addressed the challenges of class imbalance by applying class weighting, although the use of more sophisticated techniques, such as focal loss, is recommended for future work. This research highlights the potential of transformer-based models for aviation safety classification and provides a foundation for future research to improve model interpretability and generalizability across diverse datasets. These findings contribute to the growing body of research on applying deep learning techniques to aviation safety and underscore opportunities for further exploration. Full article

Hip flexion range of motion (ROM) during gait is an important surgery outcome for patients undergoing total hip arthroplasty (THA) that could help patient monitoring and rehabilitation. To allow systematic measurements during patients’ clinical pathways, hip ROM measurement should be as simple and cheap as possible to ensure patient and clinician acceptance. Single IMU options can match these requirements and offer measurements both during daily living conditions and standardized clinical tests (e.g., 10 m walk, timed up-and-go). However, single-IMU approaches to measure hip ROM have been limited. Thus, the objective of this study was to explore the accuracy of one IMU in measuring hip ROM during gait and to determine whether a single-IMU approach can provide results comparable to those of multi-IMU systems. To assess this, machine learning models were employed, ranging from the simplest (linear regression) to more complex approaches (artificial neural networks). Eighteen patients undergoing THA and seven controls were measured using a 3D opto-electronic motion capture system and one thigh-mounted IMU. Hip ROM was predicted from thigh ROM using regression and classification models and was compared to the reference hip ROM. Multiple regression was the best-performing model, with limits of agreement (LoA) of ±13° and a systematic bias of 0. Random forest, RNN, GRU and LSTM models yielded LoA ranges > 27.8°, exceeding the threshold of acceptable error. These results showed that one IMU can measure hip ROM with errors comparable to those of two-IMU methods, with potential for improvement. Using multiple linear regression was sufficient and more appropriate than employing complex ANN models. This approach offers simplicity and acceptance to users in clinical settings. Full article

With the rapid growth in social network comments, the need for more effective methods to classify their polarity—negative, neutral, or positive—has become essential. Sentiment analysis, powered by natural language processing, has evolved significantly with the adoption of advanced deep learning techniques. Long Short-Term Memory networks capture long-range dependencies in text, while transformers, with their attention mechanisms, excel at preserving contextual meaning and handling high-dimensional, semantically complex data. This study compares the performance of sentiment analysis models based on LSTM and BERT architectures using key evaluation metrics. The dataset consists of business reviews from the Yelp Open Dataset. We tested LSTM-based methods against BERT and its variants—RoBERTa, BERTweet, and DistilBERT—leveraging popular pipelines from the Hugging Face Hub. A class-by-class performance analysis is presented, revealing that more complex BERT-based models do not always guarantee superior results in the classification of Yelp reviews. Additionally, the use of bidirectionality in LSTMs does not necessarily lead to better performance. However, across a diversity of test sets, transformer models outperform traditional RNN-based models, as their generalization capability is greater than that of a simple LSTM model. Full article

This research presents a hybrid approach of Long Short-Term Memory (LSTM) and Support Vector Machine (SVM) model for greenhouse environmental monitoring, integrating machine learning and Internet of Things (IoT)-based sensing to enhance climate prediction and classification. Unlike traditional single-method approaches, this dual-model system provides a comprehensive framework for real-time climate control, optimizing temperature and humidity forecasting while enabling accurate weather classification. The LSTM model excels in capturing sequential patterns, achieving superior temperature prediction performance with a Root-Mean-Square Error (RMSE) of 0.0766, Mean Absolute Error (MAE) of 0.0454, and coefficient of determination (R²) of 0.8825. For humidity forecasting, our comparative analysis revealed that the Simple Recurrent Neural Network (RNN) demonstrates the best accuracy (RMSE: 5.3034, MAE: 3.8041, R²: 0.8187), an unexpected finding that highlights the importance of parameter-specific model selection. Simultaneously, the SVM model classifies environmental states with an accuracy of 0.63, surpassing traditional classifiers such as Logistic Regression and K Nearest Neighbors (KNN). To enhance real-time data collection and transmission, the ESP NOW wireless protocol is integrated, ensuring low latency and reliable communication between greenhouse sensors. The proposed hybrid LSTM-SVM system, combined with IoT technology, represents a significant advancement in proactive greenhouse management, offering a scalable and sustainable solution for optimizing plant growth, resource allocation, and climate adaptation. Full article

In order to study an algorithm that recognizes the state of health (SOH) of a battery rapidly and can be easily integrated into the micro-controller unit (MCU), it is proposed that signal temporal logic (STL) language is employed to describe the discharge curves, because the STL language is a formal language with strict mathematical definitions and the syntax is composed of simple logic, “and”, “or”, and “not”, under the constraints of time and parameter variation ranges, which is realizable and interpretable. Firstly, the drop voltage amplitude, drop time, voltage rebound amplitude, voltage rebound time, starting voltage, and ending voltage of the discharge curve are selected as the features of the STL formula, so the first-level and second-level primitive formulas are constructed to express the voltage of a battery in good health and poor health clearly. Secondly, the impurity measures of the information gain, misclassification gain, Gini gain, and robust extended gain are presented as the objective functions. Thirdly, the interpreter embedded in the MCU can interpret and execute each STL sentence. The voltage of a battery in good health rises slowly and falls slowly, while the voltage of a battery in poor health rises quickly and falls quickly. When the STL describes the discharge curve as “slow down slow up”, the battery is in good health. When the STL describes the discharge curve as “fast down, fast up”, the battery is in poor health. Among the different objective functions, the highest mean accuracy of the STL reaches 87.5%. In terms of the mean runtime, the extended misclassification gain and the extended Gini gain of the first-level primitives are 00851s and 0.0993, respectively. Under the same mean accuracy of 87%, the information gain and Gini gain of the second-level primitives are 0.2593 s and 0.2341 s. Compared with the existing machine learning algorithms, in terms of the mean runtime, the STL algorithm is superior to the CNN-BiLSTM-MHA model, RNN-LSTM-GRU model, and EC-MKRVM model. In terms of the mean accuracy, compared with the highest correct rate of the CNN-BiLSTM-MHA model, that is, 91.7%, the difference is 4%. As a means of quickly detecting whether the battery is in a healthy state, the accuracy difference is negligible, so the STL algorithm is apparently superior in terms of performance and realizability. Full article

Tool wear monitoring is crucial for optimizing cutting performance, reducing costs, and improving production efficiency. Existing tool wear prediction models usually design integrated models based on a convolutional neural network (CNN) and recurrent neural network (RNN) to extract spatial and temporal features separately. However, the topological structures between multi-sensor networks are ignored, and the ability to extract spatial features is limited. To overcome these limitations, a novel spatial–temporal adaptive graph convolutional network (STAGCN) is proposed to capture spatial–temporal dependencies with multi-sensor signals. First, a simple linear model is used to capture temporal patterns in individual time-series data. Second, a spatial–temporal layer composed of a bidirectional Mamba and an adaptive graph convolution is established to extract degradation features and reflect the dynamic degradation trend using an adaptive graph. Third, multi-scale triple linear attention (MTLA) is used to fuse the extracted multi-scale features across spatial, temporal, and channel dimensions, which can assign different weights adaptively to retain important information and weaken the influence of redundant features. Finally, the fused features are fed into a linear regression layer to estimate the tool wear. Experimental results conducted on the PHM2010 dataset demonstrate the effectiveness of the proposed STAGCN model, achieving a mean absolute error (MAE) of 3.40 μm and a root mean square error (RMSE) of 4.32 μm in the average results across three datasets. Full article

With the continuous advancement of deep learning algorithms and the rapid growth of computational resources, deep learning technology has undergone numerous milestone developments, evolving from simple BP neural networks into more complex and powerful network models such as CNNs, LSTMs, RNNs, and GANs. In recent years, the application of deep learning technology in ionospheric modeling has achieved breakthrough advancements, significantly impacting navigation, communication, and space weather forecasting. Nevertheless, due to limitations in observational networks and the dynamic complexity of the ionosphere, deep learning-based ionospheric models still face challenges in terms of accuracy, resolution, and interpretability. This paper systematically reviews the development of deep learning applications in ionospheric modeling, summarizing findings that demonstrate how integrating multi-source data and employing multi-model ensemble strategies has substantially improved the stability of spatiotemporal predictions, especially in handling complex space weather events. Additionally, this study explores the potential of deep learning in ionospheric modeling for the early warning of geological hazards such as earthquakes, volcanic eruptions, and tsunamis, offering new insights for constructing ionospheric-geological activity warning models. Looking ahead, research will focus on developing hybrid models that integrate physical modeling with deep learning, exploring adaptive learning algorithms and multi-modal data fusion techniques to enhance long-term predictive capabilities, particularly in addressing the impact of climate change on the ionosphere. Overall, deep learning provides a powerful tool for ionospheric modeling and indicates promising prospects for its application in early warning systems and future research. Full article

This paper presents an analysis of stock price forecasting in the financial market, with an emphasis on approaches based on time series models and deep learning techniques. Fundamental concepts of technical analysis are explored, such as exponential and simple averages, and various global indices are analyzed to be used as inputs for machine learning models, including Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Convolutional Neural Network (CNN), and XGBoost. The results show that while each model possesses distinct characteristics, selecting the most efficient approach heavily depends on the specific data and forecasting objectives. The complexity of advanced models such as XGBoost and GRU is reflected in their overall performance, suggesting that they can be particularly effective at capturing patterns and making accurate predictions in more complex time series, such as stock prices. Full article

In the current competition process of e-commerce platforms, the technical and algorithmic wars that can quickly grasp user needs and accurately recommend target commodities are the core tools of platform competition. At the same time, the existing online purchase behavior prediction models lack consideration of time series features. This paper combines the Recurrent Neural Network, which is more suitable for the commodity recommendation scenario of the e-commerce platform, with Naive Bayes, which is simple in logic and efficient in operation and constructs the online purchase behavior prediction model RNN-NB, which can consider the features of time series. The RNN-NB model is trained and tested using 3 million time series data with purchase behavior provided by the Ali Tianchi big data platform. The prediction effect of the RNN-NB model and Naive Bayes model is evaluated and compared respectively under the same experimental conditions. The results show that the overall prediction effect of the RNN-NB model is better and more stable. In addition, through the analysis of user time series features, it can be found that the possibility of user purchase is negatively correlated with the length of time series, and merchants should pay more attention to those users with shorter time series in commodity recommendation and targeted offers. The contributions of this paper are as follows: (1) By constructing an online purchasing behavior model RNN-NB, which integrates the N vs 1 structure Recurrent Neural Network and naive Bayesian model, the validity limitations of some single-architecture recommendation algorithms are solved. (2) Based on the existing naive Bayesian model, the prediction accuracy of online purchasing behavior is further improved. (3) The analysis based on the features of the time series provides new ideas for the research of later scholars and new guidance for the marketing of platform merchants. Full article

Credit card fraud detection is a critical challenge in the financial industry, with substantial economic implications. Conventional machine learning (ML) techniques often fail to adapt to evolving fraud patterns and underperform with imbalanced datasets. This study proposes a hybrid deep learning framework that integrates Generative Adversarial Networks (GANs) with Recurrent Neural Networks (RNNs) to enhance fraud detection capabilities. The GAN component generates realistic synthetic fraudulent transactions, addressing data imbalance and enhancing the training set. The discriminator, implemented using various DL architectures, including Simple RNN, Long Short-Term Memory (LSTM) networks, and Gated Recurrent Units (GRUs), is trained to distinguish between real and synthetic transactions and further fine-tuned to classify transactions as fraudulent or legitimate. Experimental results demonstrate significant improvements over traditional methods, with the GAN-GRU model achieving a sensitivity of 0.992 and specificity of 1.000 on the European credit card dataset. This work highlights the potential of GANs combined with deep learning architectures to provide a more effective and adaptable solution for credit card fraud detection. Full article

Polymer materials have garnered significant attention due to their exceptional mechanical properties and diverse industrial applications. Understanding the glass transition temperature (T_g) of polymers is critical to prevent operational failures at specific temperatures. Traditional methods for measuring T_g, such as differential scanning calorimetry (DSC) and dynamic mechanical analysis, while accurate, are often time-consuming, costly, and susceptible to inaccuracies due to random and uncertain factors. To address these limitations, the aim of the present study is to investigate the potential of Simplified Molecular Input Line Entry System (SMILES) as descriptors in simple machine learning models to predict T_g efficiently and reliably. Five models were utilized: k-nearest neighbors (KNNs), support vector regression (SVR), extreme gradient boosting (XGBoost), artificial neural network (ANN), and recurrent neural network (RNN). SMILES descriptors were converted into numerical data using either One Hot Encoding (OHE) or Natural Language Processing (NLP). The study found that SMILES inputs with fewer than 200 characters were inadequate for accurately describing compound structures, while inputs exceeding 200 characters diminished model performance due to the curse of dimensionality. The ANN model achieved the highest R² value of 0.79; however, the XGB model, with an R² value of 0.774, exhibited the highest stability and shorter training times compared to other models, making it the preferred choice for T_g prediction. The efficiency of the OHE method over NLP was demonstrated by faster training times across the KNN, SVR, XGB, and ANN models. Validation of new polymer data showed the XGB model’s robustness, with an average prediction deviation of 9.76 from actual T_g values. These findings underscore the importance of optimizing SMILES conversion methods and model parameters to enhance prediction reliability. Future research should focus on improving model accuracy and generalizability by incorporating additional features and advanced techniques. This study contributes to the development of efficient and reliable predictive models for polymer properties, facilitating the design and application of new polymer materials. Full article

The transition from internal combustion engine vehicles to electric vehicles (EVs) is gaining momentum due to their significant environmental and economic benefits. This study addresses the challenges of integrating renewable energy sources, particularly solar power, into EV charging infrastructures by using deep learning models to predict photovoltaic (PV) power generation and EV charging demand. The study determines the optimal battery energy storage capacity and charging schedule based on the prediction result and actual data. A dataset of a 15 kWp rooftop PV system and simulated EV charging data are used. The results show that simple RNNs are most effective at predicting PV power due to their adept handling of simple patterns, while bidirectional LSTMs excel at predicting EV charging demand by capturing complex dynamics. The study also identifies an optimal battery storage capacity that will balance the use of the grid and surplus solar power through strategic charging scheduling, thereby improving the sustainability and efficiency of solar energy in EV charging infrastructures. This research highlights the potential for integrating renewable energy sources with advanced energy storage solutions to support the growing electric vehicle infrastructure. Full article

Bringing out brain activity through the interpretation of EEG signals is a challenging problem that involves combined methods of signal analysis. The issue of classifying mental states induced by arithmetic tasks can be solved through various classification methods, using diverse characteristic parameters of EEG signals in the time, frequency, and statistical domains. This paper explores the results of an experiment that aimed to highlight arithmetic mental tasks contained in the PhysioNet database, performed on a group of 36 subjects. The majority of publications on this topic deal with machine learning (ML)-based classification methods with supervised learning support vector machine (SVM) algorithms, K-Nearest Neighbor (KNN), Linear Discriminant Analysis (LDA), and Decision Trees (DTs). Also, there are frequent approaches based on the analysis of EEG data as time series and their classification with Recurrent Neural Networks (RNNs), as well as with improved algorithms such as Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (BLSTM), and Gated Recurrent Units (GRUs). In the present work, we evaluate the classification method based on the comparison of domain limits for two specific characteristics of EEG signals: the statistical correlation of pairs of signals and the size of the spectral peak detected in theta, alpha, and beta bands. This study provides some interpretations regarding the electrical activity of the brain, consolidating and complementing the results of similar research. The classification method used is simple and easy to apply and interpret. The analysis of EEG data showed that the theta and beta frequency bands were the only discriminators between the relaxation and arithmetic calculation states. Notably, the F7 signal, which used the spectral peak criterion, achieved the best classification accuracy (100%) in both theta and beta bands for the subjects with the best results in performing calculations. Also, our study found the Fz signal to be a good sensor in the theta band for mental task discrimination for all subjects in the group with 90% accuracy. Full article

Objective: To train a machine-learning (ML) algorithm to classify stumbling in transfemoral amputee gait. Methods: Three subjects completed gait trials in which they were induced to stumble via three different means. Several iterations of ML algorithms were developed to ultimately classify whether individual steps were stumbles or normal gait using leave-one-out methodology. Data cleaning and hyperparameter tuning were applied. Results: One hundred thirty individual stumbles were marked and collected during the trials. Single-layer networks including Long-Short Term Memory (LSTM), Simple Recurrent Neural Network (SimpleRNN), and Gradient Recurrent Unit (GRU) were evaluated at 76% accuracy (LSTM and GRU). A four-layer LSTM achieved an 88.7% classic accuracy, with 66.9% step-specific accuracy. Conclusion: This initial trial demonstrated the ML capabilities of the gathered dataset. Though further data collection and exploration would likely improve results, the initial findings demonstrate that three forms of induced stumble can be learned with some accuracy. Significance: Other datasets and studies, such as that of Chereshnev et al. with HuGaDB, demonstrate the cataloging of human gait activities and classifying them for activity prediction. This study suggests that the integration of stumble data with such datasets would allow a knee prosthesis to detect stumbles and adapt to gait activities with some accuracy without depending on state-based recognition. Full article