Multi-View Temporal Ensemble for Classification of Non-Stationary Signals

Koh, Bee Hock David; Woo, Wai Lok

doi:10.1109/access.2019.2903571

Cited by 16 publications

(14 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As a future work, the performance of the sleep quality prediction model can be optimized by integrating swarm-based optimization techniques such as particle swarm optimization, elephant search algorithm and wolf search algorithm [62] with deep learning. Also, techniques of fusion of complementary lifestyle features with the sleep features to improve the sleep quality prediction accuracy can be seen as a future work [36,63]. Field programmable gate array-based implementation of deep learning techniques [64] for real time sleep quality prediction can be implemented within smartwatches as a future work to avoid overhead of collection and storage of data before feeding into the prediction model.…”

Section: Sleep Quality Prediction Using Sleepconsmentioning

confidence: 99%

Analysis of Data from Wearable Sensors for Sleep Quality Estimation and Prediction Using Deep Learning

2020

View full text Add to dashboard Cite

Wearable devices such as smartwatches, wristbands, GPS shoes are increasingly used for fitness and wellness as they allow users to monitor their daily health. These devices have sensors for accumulating user activity data. Clinical actigraph devices fall in the category of wearable devices worn on the wrist determined to estimate sleep parameters by recording movements during sleep. This study aims to predict sleep quality from wearable sensors using deep learning techniques. Three sleep indicators are proposed which are calculated using the data collected automatically from wearable devices. These sleep indicators are Daily Sleep Quality, Weekly Sleep Quality, and Sleep Consistency. Two deep learning models namely Convolution Neural Network (CNN) and Multilayer Perceptron (MLP) have been implemented to predict sleep quality on the basis of the proposed indicators. Two datasets have been used to validate the work proposed in this study which include a dataset comprising sleep parameters using commercial wearable devices and another dataset consisting of sleep data using clinical actigraph device. Systematic Minority Oversampling Technique has been applied for data augmentation with the intent to increase data instances and to alleviate class imbalance. CNN is observed to outperform MLP in predicting sleep quality with the highest accuracy of 97.30%. This study also evaluates the worth of each sleep attribute using Information Gain algorithm in order to identify the most important attributes which contribute to the sleep quality. It has been concluded that in bed awake percentage contributes maximum to the Daily Sleep Quality, average sleep efficiency contributes maximum to the Weekly Sleep Quality and standard deviation of midpoint of in bed and out of bed times contributes maximum to the Sleep Consistency.

show abstract

Section: Sleep Quality Prediction Using Sleepconsmentioning

confidence: 99%

Analysis of Data from Wearable Sensors for Sleep Quality Estimation and Prediction Using Deep Learning

2020

View full text Add to dashboard Cite

show abstract

“…Besides time series classification, deep neural networks such as pyramid recurrent neural network [ 33 ] can be used for change point detection to detect abrupt or gradual changes in the signal characteristics, achieved by transforming the time series data into a pyramid of multiscale feature maps in a trainable wavelet layer. In addition, an ensemble of neural networks can be used to boost the performance of time series classification [ 34 , 35 , 36 ]. InceptionTime [ 37 ] is one where a set of five different models formed by cascading multiple deep convolution neural networks, called the Inception module [ 38 ], are used.…”

Section: Introductionmentioning

confidence: 99%

Deep Temporal Convolution Network for Time Series Classification

Koh

Lim

Rahimi

et al. 2021

Sensors

Self Cite

View full text Add to dashboard Cite

A neural network that matches with a complex data function is likely to boost the classification performance as it is able to learn the useful aspect of the highly varying data. In this work, the temporal context of the time series data is chosen as the useful aspect of the data that is passed through the network for learning. By exploiting the compositional locality of the time series data at each level of the network, shift-invariant features can be extracted layer by layer at different time scales. The temporal context is made available to the deeper layers of the network by a set of data processing operations based on the concatenation operation. A matching learning algorithm for the revised network is described in this paper. It uses gradient routing in the backpropagation path. The framework as proposed in this work attains better generalization without overfitting the network to the data, as the weights can be pretrained appropriately. It can be used end-to-end with multivariate time series data in their raw form, without the need for manual feature crafting or data transformation. Data experiments with electroencephalogram signals and human activity signals show that with the right amount of concatenation in the deeper layers of the proposed network, it can improve the performance in signal classification.

show abstract

“…The learning weight of the full-connected network will be determined by the value from the previous associated training model. Next, the model-learning approaches characterize the nonlinear long-term scenarios to describe the multi-scale temporal and spatial relationships between landslide occurrence and various influencing factors [ 26 ]. To the best of our knowledge, based on literature reviews, Bidirectional-LSTM (Bi-LSTM) has rarely been applied for the susceptible landslide-prediction problem yet.…”

Section: Introductionmentioning

confidence: 99%

Automated Landslide-Risk Prediction Using Web GIS and Machine Learning Models

Tengtrairat

Woo

Parathai

et al. 2021

Sensors

Self Cite

View full text Add to dashboard Cite

Spatial susceptible landslide prediction is the one of the most challenging research areas which essentially concerns the safety of inhabitants. The novel geographic information web (GIW) application is proposed for dynamically predicting landslide risk in Chiang Rai, Thailand. The automated GIW system is coordinated between machine learning technologies, web technologies, and application programming interfaces (APIs). The new bidirectional long short-term memory (Bi-LSTM) algorithm is presented to forecast landslides. The proposed algorithm consists of 3 major steps, the first of which is the construction of a landslide dataset by using Quantum GIS (QGIS). The second step is to generate the landslide-risk model based on machine learning approaches. Finally, the automated landslide-risk visualization illustrates the likelihood of landslide via Google Maps on the website. Four static factors are considered for landslide-risk prediction, namely, land cover, soil properties, elevation and slope, and a single dynamic factor i.e., precipitation. Data are collected to construct a geospatial landslide database which comprises three historical landslide locations—Phu Chifa at Thoeng District, Ban Pha Duea at Mae Salong Nai, and Mai Salong Nok in Mae Fa Luang District, Chiang Rai, Thailand. Data collection is achieved using QGIS software to interpolate contour, elevation, slope degree and land cover from the Google satellite images, aerial and site survey photographs while the physiographic and rock type are on-site surveyed by experts. The state-of-the-art machine learning models have been trained i.e., linear regression (LR), artificial neural network (ANN), LSTM, and Bi-LSTM. Ablation studies have been conducted to determine the optimal parameters setting for each model. An enhancement method based on two-stage classifications has been presented to improve the landslide prediction of LSTM and Bi-LSTM models. The landslide-risk prediction performances of these models are subsequently evaluated using real-time dataset and it is shown that Bi-LSTM with Random Forest (Bi-LSTM-RF) yields the best prediction performance. Bi-LSTM-RF model has improved the landslide-risk predicting performance over LR, ANNs, LSTM, and Bi-LSTM in terms of the area under the receiver characteristic operator (AUC) scores by 0.42, 0.27, 0.46, and 0.47, respectively. Finally, an automated web GIS has been developed and it consists of software components including the trained models, rainfall API, Google API, and geodatabase. All components have been interfaced together via JavaScript and Node.js tool.

show abstract

Multi-View Temporal Ensemble for Classification of Non-Stationary Signals

Cited by 16 publications

References 29 publications

Analysis of Data from Wearable Sensors for Sleep Quality Estimation and Prediction Using Deep Learning

Analysis of Data from Wearable Sensors for Sleep Quality Estimation and Prediction Using Deep Learning

Deep Temporal Convolution Network for Time Series Classification

Automated Landslide-Risk Prediction Using Web GIS and Machine Learning Models

Contact Info

Product

Resources

About