Search

COVID-19 evolution imposes significant challenges for the European healthcare system. The heterogeneous spread of the pandemic within EU regions elicited a wide range of policies, such as school closure, transport restrictions, etc. However, the implementation of these interventions is not accompanied by the implementation of quantitative methods, which would indicate their effectiveness. As a result, the efficacy of such policies on reducing the spread of the virus varies significantly. This paper investigates the effectiveness of using deep learning paradigms to accurately model the spread of COVID-19. The deep learning approaches proposed in this paper are able to effectively map the temporal evolution of a COVID-19 outbreak, while simultaneously taking into account policy interventions directly into the modelling process. Thus, our approach facilitates data-driven decision making by utilizing previous knowledge to train models that predict not only the spread of COVID-19, but also the effect of specific policy measures on minimizing this spread. Global models at the EU level are proposed, which can be successfully applied at the national level. These models use various inputs in order to successfully model the spatio-temporal variability of the phenomenon and obtain generalization abilities. The proposed models are compared against the traditional epidemiological and Autoregressive Integrated Moving Average (ARIMA) models.

Over the past few years, positron emission tomography/computed tomography (PET/CT) imaging for computer-aided diagnosis has received increasing attention. Supervised deep learning architectures are usually employed for the detection of abnormalities, with anatomical localization, especially in the case of CT scans. However, the main limitations of the supervised learning paradigm include (i) large amounts of data required for model training, and (ii) the assumption of fixed network weights upon training completion, implying that the performance of the model cannot be further improved after training. In order to overcome these limitations, we apply a few-shot learning (FSL) scheme. Contrary to traditional deep learning practices, in FSL the model is provided with less data during training. The model then utilizes end-user feedback after training to constantly improve its performance. We integrate FSL in a U-Net architecture for lung cancer lesion segmentation on PET/CT scans, allowing for dynamic model weight fine-tuning and resulting in an online supervised learning scheme. Constant online readjustments of the model weights according to the users' feedback, increase the detection and classification accuracy, especially in cases where low detection performance is encountered. Our proposed method is validated on the Lung-PET-CT-DX TCIA database. PET/CT scans from 87 patients were included in the dataset and were acquired 60 minutes after intravenous 18 F-FDG injection. Experimental results indicate the superiority of our approach compared to other state-of-the-art methods., (© 2022 IOP Publishing Ltd.)

An interpretable deep learning framework for land use and land cover (LULC) classification in remote sensing using Shapley additive explanations (SHAPs) is introduced. It utilizes a compact convolutional neural network (CNN) model for the classification of satellite images and then feeds the results to a SHAP deep explainer so as to strengthen the classification results. The proposed framework is applied to Sentinel-2 satellite images containing 27000 images of pixel size $64 \times 64$ and operates on three-band combinations, reducing the model’s input data by 77% considering that 13 channels are available, while at the same time investigating on how different spectrum bands affect predictions on the dataset’s classes. Experimental results on the EuroSAT dataset demonstrate the CNN’s accurate classification with an overall accuracy of 94.72 %, whereas the classification accuracy on three-band combinations on each of the dataset’s classes highlights its improvement when compared to standard approaches with larger number of trainable parameters. The SHAP explainable results of the proposed framework shield the network’s predictions by showing correlation values that are relevant to the predicted class, thereby improving the classifications occurring in urban and rural areas with different land uses in the same scene. [ABSTRACT FROM AUTHOR]

In this study, we propose a tensor-based learning model to efficiently detect abnormalities on digital mammograms. Due to the fact that the availability of medical data is limited and often restricted by GDPR (general data protection regulation) compliance, the need for more sophisticated and less data-hungry approaches is urgent. Accordingly, our proposed artificial intelligence framework utilizes the canonical polyadic decomposition to decrease the trainable parameters of the wrapped Rank-R FNN model, leading to efficient learning using small amounts of data. Our model was evaluated on the open source digital mammographic database INBreast and compared with state-of-the-art models in this domain. The experimental results show that the proposed solution performs well in comparison with the other deep learning models, such as AlexNet and SqueezeNet, achieving 90% ± 4% accuracy and an F1 score of 84% ± 5%. Additionally, our framework tends to attain more robust performance with small numbers of data and is computationally lighter for inference purposes, due to the small number of trainable parameters. [ABSTRACT FROM AUTHOR]

Non-Intrusive Load Monitoring (NILM) describes the process of inferring the consumption pattern of appliances by only having access to the aggregated household signal. Sequence-to-sequence deep learning models have been firmly established as state-of-the-art approaches for NILM, in an attempt to identify the pattern of the appliance power consumption signal into the aggregated power signal. Exceeding the limitations of recurrent models that have been widely used in sequential modeling, this paper proposes a transformer-based architecture for NILM. Our approach, called ELECTRIcity, utilizes transformer layers to accurately estimate the power signal of domestic appliances by relying entirely on attention mechanisms to extract global dependencies between the aggregate and the domestic appliance signals. Another additive value of the proposed model is that ELECTRIcity works with minimal dataset pre-processing and without requiring data balancing. Furthermore, ELECTRIcity introduces an efficient training routine compared to other traditional transformer-based architectures. According to this routine, ELECTRIcity splits model training into unsupervised pre-training and downstream task fine-tuning, which yields performance increases in both predictive accuracy and training time decrease. Experimental results indicate ELECTRIcity's superiority compared to several state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Man overboard is an emergency in which fast and efficient detection of the critical event is the key factor for the recovery of the victim. Its severity urges the utilization of intelligent video surveillance systems that monitor the ship's perimeter in real time and trigger the relative alarms that initiate the rescue mission. In terms of deep learning analysis, since man overboard incidents occur rarely, they present a severe class imbalance problem, and thus, supervised classification methods are not suitable. To tackle this obstacle, we follow an alternative philosophy and present a novel deep learning framework that formulates man overboard identification as an anomaly detection task. The proposed system, in the absence of training data, utilizes a multi-property spatiotemporal convolutional autoencoder that is trained only on the normal situation. We explore the use of RGB video sequences to extract specific properties of the scene, such as gradient and saliency, and utilize the autoencoders to detect anomalies. To the best of our knowledge, this is the first time that man overboard detection is made in a fully unsupervised manner while jointly learning the spatiotemporal features from RGB video streams. The algorithm achieved 97.30% accuracy and a 96.01% F1-score, surpassing the other state-of-the-art approaches significantly. [ABSTRACT FROM AUTHOR]

Modeling ionospheric variability throughout a proper total electron content (TEC) parameter estimation is a demanding, however, crucial, process for achieving better accuracy and rapid convergence in precise point positioning (PPP). In particular, the single-frequency PPP (SF-PPP) method lacks accuracy due to the difficulty of dealing adequately with the ionospheric error sources. In order to apply ionosphere corrections in techniques, such as SF-PPP, external information of global ionosphere maps (GIMs) is crucial. In this article, we propose a deep learning model to efficiently predict TEC values and to replace the GIM-derived data that inherently have a global character, with equal or better in accuracy regional ones. The proposed model is suitable for predicting the ionosphere delay at different locations of receiver stations. The model is tested during different periods of time, under different solar and geomagnetic conditions and for stations in various latitudes, providing robust estimations of the ionospheric activity at the regional level. Our proposed model is a hybrid model comprising of a 1-D convolutional layer used for the optimal feature extraction and stacked recurrent layers used for temporal time series modeling. Thus, the model achieves good performance in TEC modeling compared to other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

We present a S imultaneous P recise L ocalization A nd C lassification architecture employing a U-Net model (called SPLAC U-Net) for an automated simultaneous detection/localization of corrosion and recognition of rust grade from RGB images on metal structures. The proposed architecture is compatible with the innovative idea of automated maintenance and prefabrication since (i) rust detection is carried out at pixel level accuracy in a highly precise manner and (ii) localization/classification is performed simultaneously. Prior works on corrosion identification tend to focus either on localizing defected regions with bounding boxes, with low therefore localization precision, or on only visual inspection or classification and thus they are incompatible with the prefabrication concept. The proposed scheme consists of three layers. In the first layer, we utilize a lightweight deep U-Net model for semantically localizing corroded regions in images. In the second layer, we adopt a novel data projection scheme for refining the contours of the detected objects (i.e., defected regions) of the first layer, enhancing, therefore, localization precision. We find that the network after the boundary refinement process exceeds the typical deep learning methods performance on the accuracy, F1-score, and Intersection over Union (IoU) metrics. In the third and final layer of SPLAC U-Net, we extend the U-Net model by adding residual attention modules to efficiently classify multiple rust grades based on the ISO 8501-1 taxonomy. Experimental results using objective evaluation criteria, such as accuracy, precision, recall, F1-score, IoU, and Receiver Operating Characteristic Curve (ROC) show the outperformance of the proposed SLPAC U-Net in simultaneous detecting/localizing and classifying rust regions compared to other state-of-the-art methods exploiting either color distribution properties or traditional deep learning models. Lastly, a comparison of the computational performance shows that the total required time of the proposed SPLAC U-Net model (segmentation + projection refinement) is less than solely segmenting an image by using traditional deep learning models. • Corrosion localization and rust grade classification based on the ISO 8501-1 taxonomy, using a 3-layered U-Net architecture. • Projection-based boundary refinement to promote robotic-driven maintenance and prefabrication. • Comparison with color distribution methods shows an error reduction of 37% (F1-score) and 32% (IoU) in localizing corrosion. • Comparison with other deep learning methods shows an error reduction of 14% (F1-score) and 13% (IoU) in localizing corrosion. • Comparison with a conventional U-Net shows an error reduction of 28% (F1-score) and 22% (IoU) in classifying rust grades. [ABSTRACT FROM AUTHOR]

Detecting and localizing buildings is of primary importance in urban planning tasks. Automating the building extraction process, however, has become attractive given the dominance of Convolutional Neural Networks (CNNs) in image classification tasks. In this work, we explore the effectiveness of the CNN-based architecture U-Net and its variations, namely, the Residual U-Net, the Attention U-Net, and the Attention Residual U-Net, in automatic building extraction. We showcase their robustness in feature extraction and information processing using exclusively RGB images, as they are a low-cost alternative to multi-spectral and LiDAR ones, selected from the SpaceNet 1 dataset. The experimental results show that U-Net achieves a 91.9 % accuracy, whereas introducing residual blocks, attention gates, or a combination of both improves the accuracy of the vanilla U-Net to 93.6 % , 94.0 % , and 93.7 % , respectively. Finally, the comparison between U-Net architectures and typical deep learning approaches from the literature highlights their increased performance in accurate building localization around corners and edges. [ABSTRACT FROM AUTHOR]