Loading...
17 results
Search Results
Now showing 1 - 10 of 17
- A method for sleep quality analysis based on CNN ensemble with implementation in a portable wireless devicePublication . Mendonça, Fábio; Mostafa, Sheikh Shanawaz; Dias, Fernando Morgado; Julia-Serda, Gabriel; Ravelo-Garcia, Antonio G.The quality of sleep can be affected by the occurrence of a sleep related disorder and, among these disorders, obstructive sleep apnea is commonly undiagnosed. Polysomnography is considered to be the gold standard for sleep analysis. However, it is an expensive and labor-intensive exam that is unavailable to a large group of the world population. To address these issues, the main goal of this work was to develop an automatic scoring algorithm to analyze the single-lead electrocardiogram signal, performing a minute-by-minute and an overall estimation of both quality of sleep and obstructive sleep apnea. The method employs a cross-spectral coherence technique which produces a spectrographic image that fed three one-dimensional convolutional neural networks for the classification ensemble. The predicted quality of sleep was based on the electroencephalogram cyclic alternating pattern rate, a sleep stability metric. Two methods were developed to indirectly evaluate this metric, creating two sleep quality predictions that were combined with the sleep apnea diagnosis to achieve the final global sleep quality estimation. It was verified that the quality of sleep of the nineteen tested subjects was correctly identified by the proposed model, advocating the significance of clinical analysis. The model was implemented in a non-invasive and simple to self-assemble device, producing a tool that can estimate the quality of sleep and diagnose the obstructive sleep apnea at the patient’s home without requiring the attendance of a specialized technician. Therefore, increasing the accessibility of the population to sleep analysis.
- On the Use of Transformer-Based Models for Intent Detection Using Clustering AlgorithmsPublication . Moura, André; Lima, Pedro; Mendonça, Fábio; Mostafa, Sheikh Shanawaz; Dias, Fernando MorgadoChatbots are becoming increasingly popular and require the ability to interpret natural language to provide clear communication with humans. To achieve this, intent detection is cru cial. However, current applications typically need a significant amount of annotated data, which is time-consuming and expensive to acquire. This article assesses the effectiveness of different text representations for annotating unlabeled dialog data through a pipeline that examines both classical approaches and pre-trained transformer models for word embedding. The resulting embeddings were then used to create sentence embeddings through pooling, followed by dimensionality re duction, before being fed into a clustering algorithm to determine the user’s intents. Therefore, various pooling, dimension reduction, and clustering algorithms were evaluated to determine the most appropriate approach. The evaluation dataset contains a variety of user intents across differ ent domains, with varying intent taxonomies within the same domain. Results demonstrate that transformer-based models perform better text representation than classical approaches. However, combining several clustering algorithms and embeddings from dissimilar origins through ensemble clustering considerably improves the final clustering solution. Additionally, applying the uniform manifold approximation and projection algorithm for dimension reduction can substantially improve performance (up to 20%) while using a much smaller representation.
- Automated Aviation Wind Nowcasting: Exploring Feature-Based Machine Learning MethodsPublication . Alves, Décio; Mendonça, Fábio; Mostafa, Sheikh Shanawaz; Dias, Fernando MorgadoWind factors significantly influence air travel, and extreme conditions can cause operational disruptions. Machine learning approaches are emerging as a valuable tool for predicting wind pat terns. This research, using Madeira International Airport as a case study, delves into the effectiveness of feature creation and selection for wind nowcasting, focusing on predicting wind speed, direction, and gusts. Data from four sensors provided 56 features to forecast wind conditions over intervals of 2, 10, and 20 min. Five feature selection techniques were analyzed, namely mRMR, PCA, RFECV, GA, and XGBoost. The results indicate that combining new wind features with optimized feature selection can boost prediction accuracy and computational efficiency. A strong spatial correlation was observed among sensors at different locations, suggesting that the spatial-temporal context enhances predictions. The best accuracy for wind speed forecasts yielded a mean absolute percentage error of 0.35%, 0.53%, and 0.63% for the three time intervals, respectively. Wind gust errors were 0.24%, 0.33%, and 0.38%, respectively, while wind direction predictions remained challenging with errors above 100% for all intervals.
- Multiple Time Series Fusion Based on LSTM: An Application to CAP A Phase Classification Using EEGPublication . Mendonça, Fábio; Mostafa, Sheikh Shanawaz; Freitas, Diogo; Dias, Fernando Morgado; Ravelo-García, Antonio G.The Cyclic Alternating Pattern (CAP) is a periodic activity detected in the electroencephalo gram (EEG) signals. This pattern was identified as a marker of unstable sleep with several possible clinical applications; however, there is a need to develop automatic methodologies to facilitate real-world applications based on CAP assessment. Therefore, a deep learning-based EEG channels’ feature level fusion was proposed in this work and employed for the CAP A phase classification. Two optimization algorithms optimized the channel selection, fusion, and classification procedures. The developed methodologies were evaluated by fusing the information from multiple EEG channels for patients with nocturnal frontal lobe epilepsy and patients without neurological disorders. Results showed that both optimization algorithms selected a comparable structure with similar feature level fusion, consisting of three electroencephalogram channels (Fp2–F4, C4–A1, F4–C4), which is in line with the CAP protocol to ensure multiple channels’ arousals for CAP detection. Moreover, the two optimized models reached an area under the receiver operating characteristic curve of 0.82, with average accuracy ranging from 77% to 79%, a result in the upper range of the specialist agreement and best state-of-the-art works, despite a challenging dataset. The proposed methodology also has the advantage of providing a fully automatic analysis without requiring any manual procedure. Ultimately, the models were revealed to be noise-resistant and resilient to multiple channel loss, being thus suitable for real-world application.
- Heuristic Optimization of Deep and Shallow Classifiers: An Application for Electroencephalogram Cyclic Alternating Pattern DetectionPublication . Mendonça, Fábio; Mostafa, Sheikh Shanawaz; Freitas, Diogo; Dias, Fernando Morgado; Ravelo-García, Antonio G.Methodologies for automatic non-rapid eye movement and cyclic alternating pattern analysis were proposed to examine the signal from one electroencephalogram monopolar derivation for the A phase, cyclic alternating pattern cycles, and cyclic alternating pattern rate assessments. A population composed of subjects free of neurological disorders and subjects diagnosed with sleep-disordered breathing was studied. Parallel classifications were performed for non-rapid eye movement and A phase estimations, examining a one-dimension convolutional neural network (fed with the electroencephalogram signal), a long short-term memory (fed with the electroencephalogram signal or with proposed features), and a feed-forward neural network (fed with proposed features), along with a finite state machine for the cyclic alternating pattern cycle scoring. Two hyper-parameter tuning algorithms were developed to optimize the classifiers. The model with long short-term memory fed with proposed features was found to be the best, with accuracy and area under the receiver operating characteristic curve of 83% and 0.88, respectively, for the A phase classification, while for the non-rapid eye movement estimation, the results were 88% and 0.95, respectively. The cyclic alternating pattern cycle classification accuracy was 79% for the same model, while the cyclic alternating pattern rate percentage error was 22%.
- Novel Comparative Study for the Detection of COVID-19 Using CT Scan and Chest X-ray ImagesPublication . Hayat, Ahatsham; Baglat, Preety; Mendonça, Fábio; Mostafa, Sheikh Shanawaz; Dias, Fernando MorgadoThe number of coronavirus disease (COVID-19) cases is constantly rising as the pandemic continues, with new variants constantly emerging. Therefore, to prevent the virus from spreading, coronavirus cases must be diagnosed as soon as possible. The COVID-19 pandemic has had a devastating impact on people’s health and the economy worldwide. For COVID-19 detection, reverse transcription-polymerase chain reaction testing is the benchmark. However, this test takes a long time and necessitates a lot of laboratory resources. A new trend is emerging to address these limitations regarding the use of machine learning and deep learning techniques for automatic analysis, as these can attain high diagnosis results, especially by using medical imaging techniques. However, a key question arises whether a chest computed tomography scan or chest X-ray can be used for COVID-19 detection. A total of 17,599 images were examined in this work to develop the models used to classify the occurrence of COVID-19 infection, while four different classifiers were studied. These are the convolutional neural network (proposed architecture (named, SCovNet) and Resnet18), support vector machine, and logistic regression. Out of all four models, the proposed SCoVNet architecture reached the best performance with an accuracy of almost 99% and 98% on chest computed tomography scan images and chest X-ray images, respectively.
- Sentiment Analysis in Portuguese Restaurant Reviews: Application of Transformer Models in Edge ComputingPublication . Branco, Alexandre; Parada, Daniel; Silva, Marcos; Mendonça, Fábio; Mostafa, Sheikh Shanawaz; Dias, Fernando MorgadoThis study focuses on improving sentiment analysis in restaurant reviews by leveraging transfer learning and transformer-based pre-trained models. This work evaluates the suitability of pre trained deep learning models for analyzing Natural Language Processing tasks in Portuguese. It also explores the viability of utilizing edge devices for Natural Language Processing tasks, considering their computational limitations and resource constraints. Specifically, we employ bidirectional encoder representations from transformers and robustly optimized BERT approach, two state-of-the art models, to build a sentiment review classifier. The classifier’s performance is evaluated using accuracy and area under the receiver operating characteristic curve as the primary metrics. Our results demonstrate that the classifier developed using ensemble techniques outperforms the baseline model (from 0.80 to 0.84) in accurately classifying restaurant review sentiments when three classes are considered (negative, neutral, and positive), reaching an accuracy and area under the receiver operating characteristic curve higher than 0.8 when examining a Zomato restaurant review dataset, provided for this work. This study seeks to create a model for the precise classification of Portuguese reviews into positive, negative, or neutral categories. The flexibility of deploying our model on affordable hardware platforms suggests its potential to enable real-time solutions. The deployment of the model on edge computing platforms improves accessibility in resource-constrained environments.
- Multi-objective hyperparameter optimization of convolutional neural network for obstructive sleep apnea detectionPublication . Mostafa, Sheikh Shanawaz; Mendonça, Fábio; Ravelo-García, Antonio G.; Dias, Fernando MorgadoObstructive sleep apnea (OSA) is a common sleep disorder characterized by interrupted breathing during sleep. Because of the cost, complexity, and accessibility issue related to polysomnography, the gold standard test for apnea detection, automation of the diagnostic test based on a simpler method is desired. Several signals can be used for apnea detection, such as airflow and electrocardiogram. However, the reduction of airflow normally leads to a decrease in the blood oxygen saturation level (SpO2). This signal is usually measured by a pulse oximeter, a sensor that is cheap, portable, and easy to assemble. Therefore, the SpO2 was chosen as the reference signal. Feature based classifiers with shallow neural networks have been developed to provide apnea detection using SpO2. However, two main issues arise, the need for feature creation and the selection of the more relevant features. Deep neural networks can solve these issues by employing featureless methods. Among multiple deep classifiers that have been developed, convolution neural networks (CNN) are gaining popularity. However, the selection of the CNN structure and hyperparameters are typically done by experts, where prior knowledge is essential. With these problems in mind, an algorithm for automatic structure selection and hyper parameterization of a one dimension CNN was developed to detect OSA events using only the SpO2 signal. Three different input sizes and databases were tested, and the best model achieved an average accuracy, sensitivity, and specificity of 94%, 92%, and 96%, respectively.
- The Potential of Machine Learning for Wind Speed and Direction Short-Term Forecasting: A Systematic ReviewPublication . Alves, Décio; Mendonça, Fábio; Mostafa, Sheikh Shanawaz; Dias, Fernando MorgadoWind forecasting, which is essential for numerous services and safety, has significantly improved in accuracy due to machine learning advancements. This study reviews 23 articles from 1983 to 2023 on machine learning for wind speed and direction nowcasting. The wind prediction ranged from 1 min to 1 week, with more articles at lower temporal resolutions. Most works employed neural networks, focusing recently on deep learning models. Among the reported performance metrics, the most prevalent were mean absolute error, mean squared error, and mean absolute percentage error. Considering these metrics, the mean performance of the examined works was 0.56 m/s, 1.10 m/s, and 6.72%, respectively. The results underscore the novel effectiveness of machine learning in predicting wind conditions using high-resolution time data and demonstrated that deep learning models surpassed traditional methods, improving the accuracy of wind speed and direction forecasts. Moreover, it was found that the inclusion of non-wind weather variables does not benefit the model’s overall performance. Further studies are recommended to predict both wind speed and direction using diverse spatial data points, and high-resolution data are recommended along with the usage of deep learning models.
- Automatic Detection of a Phases for CAP ClassificationPublication . Mendonça, Fábio; Fred, Ana; Mostafa, Sheikh Shanawaz; Dias, Fernando Morgado; Ravelo-García, Antonio G.The aim of this study is to develop an automatic detector of the cyclic alternating pattern by first detecting the activation phases (A phases) of this pattern, analysing the electroencephalogram during sleep, and then applying a finite state machine to implement the final classification. A public database was used to test the algorithms and a total of eleven features were analysed. Sequential feature selection was employed to select the most relevant features and a post processing procedure was used for further improvement of the classification. The classification of the A phases was produced using linear discriminant analysis and the average accuracy, sensitivity and specificity was, respectively, 75%, 78% and 74%. The cyclic alternating pattern detection accuracy was 75%. When comparing with the state of the art, the proposed method achieved the highest sensitivity but a lower accuracy since the fallowed approach was to keep the REM periods, contrary to the method that is used in the majority of the state of the art publications which leads to an increase in the overall performance. However, the approach of this work is more suitable for automatic system implementation since no alteration of the EEG data is needed.