Electricity demand prediction is crucial to ensure the operational safety and cost-efficient operation of the power system. Electricity demand has predominantly been predicted deterministically, while uncertainty analysis has been usually overlooked. To address this research gap, an integrated Neural Facebook Prophet (NFBP) model and Gaussian Kernel Density Estimation (KDE) model is proposed in this paper, as a way to obtain point and interval predictions of electricity demand, quantifying this way the uncertainty in the predictions. First, historical lagged data, created by utilizing the Partial Auto-correlation Function and Mutual Information Test, is applied to train a prediction model based on NFBP, Deep Learning (DL) as well as Statistical Models. Second, the model Prediction Errors (PE) are derived from the difference between actual and predicted values. A splitting strategy based on the mean and standard deviation of PE is proposed. Finally, electricity demand prediction intervals are obtained by applying Gaussian KDE on split PE. To verify the effectiveness of the proposed model, simulation studies are carried out for three prediction horizons on freely available datasets for the Bulimba sub-station in Southeast Queensland, Australia. Compared with DL models (Long-Short Term Memory Network and Deep Neural Network), the Root Mean Square Error of the NFBP model was reduced by 6.1% and 11.3% for 0.5-hr ahead, 22.7% and 26.3% for 6-hr ahead, and 31.8% and 29.9% for daily prediction. In addition, the Prediction Interval normalized Interval width is smaller in magnitude for the proposed NFBP-KDE model compared to other DL and Statistical models

In this article, we propose a novel flexible architecture, with different algorithmic procedures, for effective story segmentation of broadcast news from subtitle files. The proposed system exploits spatial and temporal distance, as well as sentence similarity, to classify different stories in news broadcasts. The computational algorithms which form the architecture mainly focus on each sentence's features (temporal distance, spatial distance, and semantic similarity), and are combined to build an overall classifier. The first algorithm in the architecture focuses on the segmentation task, detecting boundaries between news. The second and third algorithms identify high semantic correlation between pieces of text, whether they are consecutive in space or not. Video Text Track (VTT) subtitle files are used to evaluate the performance of the proposed approach, although any file format that includes temporal information could also be considered. These VTT files may contain text errors and inaccuracies, and the proposed algorithms have been designed to deal with noisy content.

This study proposes a new hybrid deep learning (DL) model, the called CSVR, for Global Solar Radiation (GSR) predictions by integrating Convolutional Neural Network (CNN) with Support Vector Regression (SVR) approach. First, the CNN algorithm is used to extract local patterns as well as common features that occur recurrently in time series data at different intervals. Then, the SVR is subsequently adopted to replace the fully connected CNN layers to predict the daily GSR time series data at six solar farms in Queensland, Australia. To develop the hybrid CSVR model, we adopt the most pertinent meteorological variables from Global Climate Model and Scientific Information for Landowners database. From a pool of Global Climate Models variables and ground-based observations, the optimal features are selected through a metaheuristic Feature Selection algorithm, an Atom Search Optimization method. The hyperparameters of the proposed CSVR are optimized by mean of the HyperOpt method, and the overall performance of the objective algorithm is benchmarked against eight alternative DL methods, and some of the other Machine Learning approaches (LSTM, DBN, RBF, BRF, MARS, WKNNR, GPML and M5TREE) methods. The results obtained shows that the proposed CSVR model can offer several predictive advantages over the alternative DL models, as well as the conventional ML models. Specifically, we note that the CSVR model recorded a root mean square error/mean absolute error ranging between ≈ 2.172–3.305 MJ m2/1.624–2.370 MJ m2 over the six tested solar farms compared to ≈ 2.514–3.879 MJ m2/1.939–2.866 MJ m2 from alternative ML and DL algorithms. Consistent with this predicted error, the correlation between the measured and the predicted GSR, including the Willmott’s, Nash-Sutcliffe’s coefficient and Legates & McCabe’s Index was relatively higher for the proposed CSVR model compared to other DL and Machine Learning methods for all of the study sites. Accordingly, this study advocates the merits of CSVR model to provide a viable alternative to accurately predict GSR for renewable energy exploitation, energy demand or other forecasting-based applications.

Home-based physical therapies are effective if the prescribed exercises are correctly executed and patients adhere to these routines. This is specially important for older adults who can easily forget the guidelines from therapists. Inertial Measurement Units (IMUs) are commonly used for tracking exercise execution giving information of patients’ motion data. In this work, we propose the use of Machine Learning techniques to recognize which exercise is being carried out and to assess if the recognized exercise is properly executed by using data from four IMUs placed on the person limbs. To the best of our knowledge, both tasks have never been addressed together as a unique complex task before. However, their combination is needed for the complete characterization of the performance of physical therapies. We evaluate the performance of six machine learning classifiers in three contexts: recognition and evaluation in a single classifier, recognition of correct exercises, excluding the wrongly performed exercises, and a two-stage approach that first recognizes the exercise and then evaluates it. We apply our proposal to a set of 8 exercises of the upper-and lower-limbs designed for maintaining elderly people health status. To do so, the motion of 30 volunteers were monitored with 4 IMUs. We obtain accuracies of 88 . 4 % and the 91 . 4 % in the two initial scenarios. In the third one, the recognition provides an accuracy of 96 . 2 %, whereas the exercise evaluation varies between 93 . 6 % and 100 . 0 %. This work proves the feasibility of IMUs for a complete monitoring of physical therapies in which we can get information of which exercise is being performed and its quality, as a basis for designing virtual coaches.

Shape-from-Template (SfT) solves 3D vision from a single image and a deformable 3D object model, called a template. Concretely, SfT computes registration (the correspondence between the template and the image) and reconstruction (the depth in camera frame). It constrains the object deformation to quasi-isometry. Real-time and automatic SfT represents an open problem for complex objects and imaging conditions. We present four contributions to address core unmet challenges to realise SfT with a Deep Neural Network (DNN). First, we propose a novel DNN called DeepSfT, which encodes the template in its weights and hence copes with highly complex templates. Second, we propose a semi-supervised training procedure to exploit real data. This is a practical solution to overcome the render gap that occurs when training only with simulated data. Third, we propose a geometry adaptation module to deal with different cameras at training and inference. Fourth, we combine statistical learning with physics-based reasoning. DeepSfT runs automatically and in real-time and we show with numerous experiments and an ablation study that it consistently achieves a lower 3D error than previous work. It outperforms in generalisation and achieves great performance in terms of reconstruction and registration error with wide-baseline, occlusions, illumination changes, weak texture and blur.

Abstract This work describes an end-to-end approach for real-time human action recognition from raw depth image-sequences. The proposal is based on a 3D fully convolutional neural network, named 3DFCNN, which automatically encodes spatio-temporal patterns from raw depth sequences. The described 3D-CNN allows actions classification from the spatial and temporal encoded information of depth sequences. The use of depth data ensures that action recognition is carried out protecting people’s privacy, since their identities can not be recognized from these data. The proposed 3DFCNN has been optimized to reach a good performance in terms of accuracy while working in real-time. Then, it has been evaluated and compared with other state-of-the-art systems in three widely used public datasets with different characteristics, demonstrating that 3DFCNN outperforms all the non-DNN-based state-of-the-art methods with a maximum accuracy of 83.6% and obtains results that are comparable to the DNN-based approaches, while maintaining a much lower computational cost of 1.09 seconds, what significantly increases its applicability in real-world environments.