Search results

The purpose of this paper is to ensure the stable operation of satellites in orbit and to assist ground personnel in continuously monitoring the satellite telemetry data and finding anomalies in advance, which can improve the reliability of satellite operation and prevent catastrophic losses.

This paper proposes a deep auto-encoder (DAE) satellite anomaly advance warning framework for satellite telemetry data. Firstly, this study performs grey correlation analysis, extracts important feature attributes to construct feature vectors and builds the variational auto-encoder with bidirectional long short-term memory generative adversarial network discriminator (VAE/BLGAN). Then, the Mahalanobis distance is used to measure the reconstruction score of input and output. According to the periodic characteristic of satellite operation, a dynamic threshold method based on periodic time window is proposed. Satellite health monitoring and advance warning are achieved using reconstruction scores and dynamic thresholds.

Experiment results indicate DAE methods can probe that satellite telemetry data appear abnormal, trigger a warning before the anomaly occurring and thus allow enough time for troubleshooting. This paper further verifies that the proposed VAE/BLGAN model has stronger data learning ability than other two auto-encoder models and is sensitive to satellite monitoring data.

This paper provides a DAE framework to apply in the field of satellite health monitoring and anomaly advance warning. To the best of the authors’ knowledge, this is the first paper to combine DAE methods with satellite anomaly detection, which can promote the application of artificial intelligence in spacecraft health monitoring.

Excessive tool wear is responsible for damage or breakage of the tool, workpiece, or machining center. Thus, it is crucial to examine tool conditions during the machining process to improve its useful functional life and the surface quality of the final product. AI-based tool wear prediction techniques have proven to be effective in estimating the Remaining Useful Life (RUL) of the cutting tool. However, the model prediction needs improvement in terms of accuracy.

This paper represents a methodology of fusing a feature selection technique along with state-of-the-art deep learning models. The authors have used NASA milling data sets along with vibration signals for tool wear prediction and performance analysis in 15 different fault scenarios. Multiple steps are used for the feature selection and ranking. Different Long Short-Term Memory (LSTM) approaches are used to improve the overall prediction accuracy of the model for tool wear prediction. LSTM models' performance is evaluated using R-square, Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) parameters.

The R-square accuracy of the hybrid model is consistently high and has low MAE, MAPE and RMSE values. The average R-square score values for LSTM, Bidirection, Encoder–Decoder and Hybrid LSTM are 80.43, 84.74, 94.20 and 97.85%, respectively, and corresponding average MAPE values are 23.46, 22.200, 9.5739 and 6.2124%. The hybrid model shows high accuracy as compared to the remaining LSTM models.

The low variance, Spearman Correlation Coefficient and Random Forest Regression methods are used to select the most significant feature vectors for training the miscellaneous LSTM model versions and highlight the best approach. The selected features pass to different LSTM models like Bidirectional, Encoder–Decoder and Hybrid LSTM for tool wear prediction. The Hybrid LSTM approach shows a significant improvement in tool wear prediction.

Genre classification of songs based on lyrics is a challenging task even for humans, however, state-of-the-art natural language processing has recently offered advanced solutions to this task. The purpose of this study is to advance the understanding and application of natural language processing and deep learning in the domain of music genre classification, while also contributing to the broader themes of global knowledge and communication, and sustainable preservation of cultural heritage.

The main contribution of this study is the development and evaluation of various machine and deep learning models for song genre classification. Additionally, we investigated the effect of different word embeddings, including Global Vectors for Word Representation (GloVe) and Word2Vec, on the classification performance. The tested models range from benchmarks such as logistic regression, support vector machine and random forest, to more complex neural network architectures and transformer-based models, such as recurrent neural network, long short-term memory, bidirectional long short-term memory and bidirectional encoder representations from transformers (BERT).

The authors conducted experiments on both English and multilingual data sets for genre classification. The results show that the BERT model achieved the best accuracy on the English data set, whereas cross-lingual language model pretraining based on RoBERTa (XLM-RoBERTa) performed the best on the multilingual data set. This study found that songs in the metal genre were the most accurately labeled, as their text style and topics were the most distinct from other genres. On the contrary, songs from the pop and rock genres were more challenging to differentiate. This study also compared the impact of different word embeddings on the classification task and found that models with GloVe word embeddings outperformed Word2Vec and the learning embedding layer.

This study presents the implementation, testing and comparison of various machine and deep learning models for genre classification. The results demonstrate that transformer models, including BERT, robustly optimized BERT pretraining approach, distilled bidirectional encoder representations from transformers, bidirectional and auto-regressive transformers and XLM-RoBERTa, outperformed other models.

As data centres grow in size and complexity, traditional air-cooling methods are becoming less effective and more expensive. Immersion cooling, where servers are submerged in a dielectric fluid, has emerged as a promising alternative. Ensuring reliable operations in data centre applications requires the development of an effective control framework for immersion cooling systems, which necessitates the prediction of server temperature. While deep learning-based temperature prediction models have shown effectiveness, further enhancement is needed to improve their prediction accuracy. This study aims to develop a temperature prediction model using Long Short-Term Memory (LSTM) Networks based on recursive encoder-decoder architecture.

This paper explores the use of deep learning algorithms to predict the temperature of a heater in a two-phase immersion-cooled system using NOVEC 7100. The performance of recursive-long short-term memory-encoder-decoder (R-LSTM-ED), recursive-convolutional neural network-LSTM (R-CNN-LSTM) and R-LSTM approaches are compared using mean absolute error, root mean square error, mean absolute percentage error and coefficient of determination (R²) as performance metrics. The impact of window size, sampling period and noise within training data on the performance of the model is investigated.

The R-LSTM-ED consistently outperforms the R-LSTM model by 6%, 15.8% and 12.5%, and R-CNN-LSTM model by 4%, 11% and 12.3% in all forecast ranges of 10, 30 and 60 s, respectively, averaged across all the workloads considered in the study. The optimum sampling period based on the study is found to be 2 s and the window size to be 60 s. The performance of the model deteriorates significantly as the noise level reaches 10%.

The proposed models are currently trained on data collected from an experimental setup simulating data centre loads. Future research should seek to extend the applicability of the models by incorporating time series data from immersion-cooled servers.

The proposed multivariate-recursive-prediction models are trained and tested by using real Data Centre workload traces applied to the immersion-cooled system developed in the laboratory.

This study aims to propose a learning pattern analysis method which can improve a predictive model’s performance, as well as discover hidden insights into micro-level learning pattern. Analyzing student’s learning patterns can help instructors understand how their course design or activities shape learning behaviors; depict students’ beliefs about learning and their motivation; and predict learning performance by analyzing individual students’ learning patterns. Although time-series analysis is one of the most feasible predictive methods for learning pattern analysis, literature-indicated current approaches cannot provide holistic insights about learning patterns for personalized intervention. This study identified at-risk students by micro-level learning pattern analysis and detected pattern types, especially at-risk patterns that existed in the case study. The connections among students’ learning patterns, corresponding self-regulated learning (SRL) strategies and learning performance were finally revealed.

The method used long short-term memory (LSTM)-encoder to process micro-level behavioral patterns for feature extraction and compression, thus the students’ behavior pattern information were saved into encoded series. The encoded time-series data were then used for pattern analysis and performance prediction. Time series clustering were performed to interpret the unique strength of proposed method.

Successful students showed consistent participation levels and balanced behavioral frequency distributions. The successful students also adjusted learning behaviors to meet with course requirements accordingly. The three at-risk patten types showed the low-engagement (R1) the low-interaction (R2) and the non-persistent characteristics (R3). Successful students showed more complete SRL strategies than failed students. Political Science had higher at-risk chances in all three at-risk types. Computer Science, Earth Science and Economics showed higher chances of having R3 students.

The study identified multiple learning patterns which can lead to the at-risk situation. However, more studies are needed to validate whether the same at-risk types can be found in other educational settings. In addition, this case study found the distributions of at-risk types were vary in different subjects. The relationship between subjects and at-risk types is worth further investigation.

This study found the proposed method can effectively extract micro-level behavioral information to generate better prediction outcomes and depict student’s SRL learning strategies in online learning. The authors confirm that the research in their work is original, and that all the data given in the paper are real and authentic. The study has not been submitted to peer review and not has been accepted for publishing in another journal.

This study explores whether a new machine learning method can more accurately predict the movement of stock prices.

This study presents a novel hybrid deep learning model, Residual-CNN-Seq2Seq (RCSNet), to predict the trend of stock price movement. RCSNet integrates the autoregressive integrated moving average (ARIMA) model, convolutional neural network (CNN) and the sequence-to-sequence (Seq2Seq) long–short-term memory (LSTM) model.

The hybrid model is able to forecast both linear and non-linear time-series component of stock dataset. CNN and Seq2Seq LSTMs can be effectively combined for dynamic modeling of short- and long-term-dependent patterns in non-linear time series forecast. Experimental results show that the proposed model outperforms baseline models on S&P 500 index stock dataset from January 2000 to August 2016.

This study develops the RCSNet hybrid model to tackle the challenge by combining both linear and non-linear models. New evidence has been obtained in predicting the movement of stock market prices.

Stock price prediction is a hot topic and traditional prediction methods are usually based on statistical and econometric models. However, these models are difficult to deal with nonstationary time series data. With the rapid development of the internet and the increasing popularity of social media, online news and comments often reflect investors’ emotions and attitudes toward stocks, which contains a lot of important information for predicting stock price. This paper aims to develop a stock price prediction method by taking full advantage of social media data.

This study proposes a new prediction method based on deep learning technology, which integrates traditional stock financial index variables and social media text features as inputs of the prediction model. This study uses Doc2Vec to build long text feature vectors from social media and then reduce the dimensions of the text feature vectors by stacked auto-encoder to balance the dimensions between text feature variables and stock financial index variables. Meanwhile, based on wavelet transform, the time series data of stock price is decomposed to eliminate the random noise caused by stock market fluctuation. Finally, this study uses long short-term memory model to predict the stock price.

The experiment results show that the method performs better than all three benchmark models in all kinds of evaluation indicators and can effectively predict stock price.

In this paper, this study proposes a new stock price prediction model that incorporates traditional financial features and social media text features which are derived from social media based on deep learning technology.

A cost-effective way to achieve fuel economy is to reinforce positive driving behaviour. Driving behaviour can be controlled if drivers can be alerted for behaviour that results in poor fuel economy. Fuel consumption must be tracked and monitored instantaneously rather than tracking average fuel economy for the entire trip duration. A single-step application of machine learning (ML) is not sufficient to model prediction of instantaneous fuel consumption and detection of anomalous fuel economy. The study designs an ML pipeline to track and monitor instantaneous fuel economy and detect anomalies.

This research iteratively applies different variations of a two-step ML pipeline to the driving dataset for hatchback cars. The first step addresses the problem of accurate measurement and prediction of fuel economy using time series driving data, and the second step detects abnormal fuel economy in relation to contextual information. Long short-term memory autoencoder method learns and uses the most salient features of time series data to build a regression model. The contextual anomaly is detected by following two approaches, kernel quantile estimator and one-class support vector machine. The kernel quantile estimator sets dynamic threshold for detecting anomalous behaviour. Any error beyond a threshold is classified as an anomaly. The one-class support vector machine learns training error pattern and applies the model to test data for anomaly detection. The two-step ML pipeline is further modified by replacing long short term memory autoencoder with gated recurrent network autoencoder, and the performance of both models is compared. The speed recommendations and feedback are issued to the driver based on detected anomalies for controlling aggressive behaviour.

A composite long short-term memory autoencoder was compared with gated recurrent unit autoencoder. Both models achieve prediction accuracy within a range of 98%–100% for prediction as a first step. Recall and accuracy metrics for anomaly detection using kernel quantile estimator remains within 98%–100%, whereas the one-class support vector machine approach performs within the range of 99.3%–100%.

The proposed approach does not consider socio-demographics or physiological information of drivers due to privacy concerns. However, it can be extended to correlate driver's physiological state such as fatigue, sleep and stress to correlate with driving behaviour and fuel economy. The anomaly detection approach here is limited to providing feedback to driver, it can be extended to give contextual feedback to the steering controller or throttle controller. In the future, a controller-based system can be associated with an anomaly detection approach to control the acceleration and braking action of the driver.

The suggested approach is helpful in monitoring and reinforcing fuel-economical driving behaviour among fleet drivers as per different environmental contexts. It can also be used as a training tool for improving driving efficiency for new drivers. It keeps drivers engaged positively by issuing a relevant warning for significant contextual anomalies and avoids issuing a warning for minor operational errors.

This paper contributes to the existing literature by providing an ML pipeline approach to track and monitor instantaneous fuel economy rather than relying on average fuel economy values. The approach is further extended to detect contextual driving behaviour anomalies and optimises fuel economy. The main contributions for this approach are as follows: (1) a prediction model is applied to fine-grained time series driving data to predict instantaneous fuel consumption. (2) Anomalous fuel economy is detected by comparing prediction error against a threshold and analysing error patterns based on contextual information.

Tourism spot or point-of-interest (POI) recommendation has become a common service in people's daily life. The purpose of this paper is to model users' check-in history in order to predict a set of locations that a user may soon visit.

The authors proposed a novel learning-based method, the pattern-based dual learning POI recommendation system as a solution to consider users' interests and the uniformity of popular POI patterns when making recommendations. Differing from traditional long short-term memory (LSTM), a new users’ regularity–POIs’ popularity patterns long short-term memory (UP-LSTM) model was developed to concurrently combine the behaviors of a specific user and common users.

The authors introduced the concept of dual learning for POI recommendation. Several performance evaluations were conducted on real-life mobility data sets to demonstrate the effectiveness and practicability of POI recommendations. The metrics such as hit rate, precision, recall and F-measure were used to measure the capability of ranking and precise prediction of the proposed model over all baselines. The experimental results indicated that the proposed UP-LSTM model consistently outperformed the state-of-the-art models in all metrics by a large margin.

This study contributes to the existing literature by incorporating a novel pattern–based technique to analyze how the popularity of POIs affects the next move of a particular user. Also, the authors have proposed an effective fusing scheme to boost the prediction performance in the proposed UP-LSTM model. The experimental results and discussions indicate that the combination of the user's regularity and the POIs’ popularity patterns in PDLRec could significantly enhance the performance of POI recommendation.

The research on structure function recognition mainly concentrates on identifying a specific part of academic literature and its applicability in the multidiscipline perspective. A specific part of academic literature, such as sentences, paragraphs and chapter contents are also called a level of academic literature in this paper. There are a few comparative research works on the relationship between models, disciplines and levels in the process of structure function recognition. In view of this, comparative research on structure function recognition based on deep learning has been conducted in this paper.

An experimental corpus, including the academic literature of traditional Chinese medicine, library and information science, computer science, environmental science and phytology, was constructed. Meanwhile, deep learning models such as convolutional neural networks (CNN), long and short-term memory (LSTM) and bidirectional encoder representation from transformers (BERT) were used. The comparative experiments of structure function recognition were conducted with the help of the deep learning models from the multilevel perspective.

The experimental results showed that (1) the BERT model performed best, with F1 values of 78.02, 89.41 and 94.88%, respectively at the level of sentence, paragraph and chapter content. (2) The deep learning models performed better on the academic literature of traditional Chinese medicine than on other disciplines in most cases, e.g. F1 values of CNN, LSTM and BERT, respectively arrived at 71.14, 69.96 and 78.02% at the level of sentence. (3) The deep learning models performed better at the level of chapter content than other levels, the maximum F1 values of CNN, LSTM and BERT at 91.92, 74.90 and 94.88%, respectively. Furthermore, the confusion matrix of recognition results on the academic literature was introduced to find out the reason for misrecognition.

This paper may inspire other research on structure function recognition, and provide a valuable reference for the analysis of influencing factors.