Search

The electrocardiogram (ECG) signal describes the heart’s electrical activity, allowing it to detect several health conditions, including cardiac system abnormalities and dysfunctions. Nowadays, most patient medical records are still paper-based, especially those made in past decades. The importance of collecting digitized ECGs is twofold: firstly, all medical applications can be easily implemented with an engineering approach if the ECGs are treated as signals; secondly, paper ECGs can deteriorate over time, therefore a correct evaluation of the patient’s clinical evolution is not always guaranteed. The goal of this paper is the realization of an automatic conversion algorithm from paper-based ECGs (images) to digital ECG signals. The algorithm involves a digitization process tested on an image set of 16 subjects, also with pathologies. The quantitative analysis of the digitization method is carried out by evaluating the repeatability and reproducibility of the algorithm. The digitization accuracy is evaluated both on the entire signal and on six ECG time parameters (R-R peak distance, QRS complex duration, QT interval, PQ interval, P-wave duration, and heart rate). Results demonstrate the algorithm efficiency has an average Pearson correlation coefficient of 0.94 and measurement errors of the ECG time parameters are always less than 1 mm. Due to the promising experimental results, the algorithm could be embedded into a graphical interface, becoming a measurement and collection tool for cardiologists.

Falls are unusual actions that cause a significant health risk among older people. The growing percentage of people of old age requires urgent development of fall detection and prevention systems. The emerging technology focuses on developing such systems to improve quality of life, especially for the elderly. A fall prevention system tries to predict and reduce the risk of falls. In contrast, a fall detection system observes the fall and generates a help notification to minimize the consequences of falls. A plethora of technical and review papers exist in the literature with a primary focus on fall detection. Similarly, several studies are relatively old, with a focus on wearables only, and use statistical and threshold-based approaches with a high false alarm rate. Therefore, this paper presents the latest research trends in fall detection and prevention systems using Machine Learning (ML) algorithms. It uses recent studies and analyzes datasets, age groups, ML algorithms, sensors, and location. Additionally, it provides a detailed discussion of the current trends of fall detection and prevention systems with possible future directions. This overview can help researchers understand the current systems and propose new methodologies by improving the highlighted issues. ispartof: SENSORS vol:21 issue:15 ispartof: location:Switzerland status: published

The increasing demand for smart vehicles with many sensing capabilities will escalate data traffic in vehicular networks. Meanwhile, available network resources are limited. The emergence of AI implementation in vehicular network resource allocation opens the opportunity to improve resource utilization to provide more reliable services. Accordingly, many resource allocation schemes with various machine learning algorithms have been proposed to dynamically manage and allocate network resources. This survey paper presents how machine learning is leveraged in the vehicular network resource allocation strategy. We focus our study on determining its role in the mechanism. First, we provide an analysis of how authors designed their scenarios to orchestrate the resource allocation strategy. Secondly, we classify the mechanisms based on the parameters they chose when designing the algorithms. Finally, we analyze the challenges in designing a resource allocation strategy in vehicular networks using machine learning. Therefore, a thorough understanding of how machine learning algorithms are utilized to offer a dynamic resource allocation in vehicular networks is provided in this study.

Road and sidewalk detection in urban scenarios is a challenging task because of the road imperfections and high sensor data bandwidth. Traditional free space and ground filter algorithms are not sensitive enough for small height differences. Camera-based or sensor-fusion solutions are widely used to classify drivable road from sidewalk or pavement. A LIDAR sensor contains all the necessary information from which the feature extraction can be done. Therefore, this paper focuses on LIDAR-based feature extraction. For road and sidewalk detection, the current paper presents a real-time (20 Hz+) solution. This solution can also be used for local path planning. Sidewalk edge detection is the combination of three algorithms working parallelly. To validate the result, the de facto standard benchmark dataset, KITTI, was used alongside our measurements. The data and the source code to reproduce the results are shared publicly on our GitHub repository.

Everyone should be vaccinated, but the eligibility and safety of the vaccine are always overlooked by most people. The outbreak of COVID-19 has led many countries to intensify the development and production of the COVID-19 vaccine. and some countries have even required universal vaccination against this epidemic. However, such popularization of vaccination has also exposed various flaws in vaccine management that existed in the past, and vaccinators have become more concerned about the effectiveness of their vaccinations. In this paper, we propose a blockchain-based traceable vaccine management system. First, the system uses smart contracts to store the records generated during the whole process, from vaccine production to vaccination. Second, the proposed scheme uses the Edwards-curve digital signature algorithm (EdDSA) to guarantee the security and integrity of these data. Third, the system participants can access the corresponding data according to their authority to ensure the transparency of the whole system operation process. Finally, this paper will also conduct a security analysis of the whole system to ensure that the system can resist potential attacks by criminals.

This paper explores the problem of severe multipath propagation of underwater acoustic signals in a swimming pool. The problem appeared in a study that examined a system used to signal emergency situations (i.e., pre-drowning symptoms detected by a wearable device on a pool user’s wrist) and locate the signal source. A swimming pool acoustic environment is characterized by the presence of large flat reflecting planes surrounding a small volume of water. The reflections are numerous and much stronger than in typical hydroacoustic applications. In this paper, we attempted to create a model of the swimming pool response, one that is suitable for simulation experiments with detection and localization of emergency signals. Then, we explore the possible remedies for the localization system, applied on the transmit side (waveform design) and on the receive side (receiver placement and signal processing). Finally, we present an algorithm for object localization, considering the possible reflections with a multi-hypothesis approach.

Nowadays, electric vehicles have gained great popularity due to their performance and efficiency. Investment in the development of this new technology is justified by increased consciousness of the environmental impacts caused by combustion vehicles such as greenhouse gas emissions, which have contributed to global warming as well as the depletion of non-oil renewable energy source. The lithium-ion battery is an appropriate choice for electric vehicles (EVs) due to its promising features of high voltage, high energy density, low self-discharge, and long life cycles. In this context, State of Charge (SoC) is one of the vital parameters of the battery management system (BMS). Nevertheless, because the discharge and charging of battery cells requires complicated chemical operations, it is therefore hard to determine the state of charge of the battery cell. This paper analyses the application of Artificial Neural Networks (ANNs) in the estimation of the SoC of lithium batteries using the NASA’s research center dataset. Normally, the learning of these networks is performed by some method based on a gradient, having the mean squared error as a cost function. This paper evaluates the substitution of this traditional function by a measure of similarity of the Information Theory, called the Maximum Correntropy Criterion (MCC). This measure of similarity allows statistical moments of a higher order to be considered during the training process. For this reason, it becomes more appropriate for non-Gaussian error distributions and makes training less sensitive to the presence of outliers. However, this can only be achieved by properly adjusting the width of the Gaussian kernel of the correntropy. The proper tuning of this parameter is done using adaptive strategies and genetic algorithms. The proposed identification model was developed using information for training and validation, using a dataset made available in a online repository maintained by NASA’s research center. The obtained results demonstrate that the use of correntropy, as a cost function in the error backpropagation algorithm, makes the identification procedure using ANN networks more robust when compared to the traditional Mean Squared Error.

Studies have shown that ordinary color cameras can detect the subtle color changes of the skin caused by the heartbeat cycle. Therefore, cameras can be used to remotely monitor the pulse in a non-contact manner. The technology for non-contact physiological measurement in this way is called remote photoplethysmography (rPPG). Heart rate variability (HRV) analysis, as a very important physiological feature, requires us to be able to accurately recover the peak time locations of the rPPG signal. This paper proposes an efficient spatiotemporal attention network (ESA-rPPGNet) to recover high-quality rPPG signal for heart rate variability analysis. First, 3D depth-wise separable convolution and a structure based on mobilenet v3 are used to greatly reduce the time complexity of the network. Next, a lightweight attention block called 3D shuffle attention (3D-SA), which integrates spatial attention and channel attention, is designed to enable the network to effectively capture inter-channel dependencies and pixel-level dependencies. Moreover, ConvGRU is introduced to further improve the network’s ability to learn long-term spatiotemporal feature information. Compared with existing methods, the experimental results show that the method proposed in this paper has better performance and robustness on the remote HRV analysis.

Hypovolemia is a physiological state of reduced blood volume that can exist as either (1) absolute hypovolemia because of a lower circulating blood (plasma) volume for a given vascular space (dehydration, hemorrhage) or (2) relative hypovolemia resulting from an expanded vascular space (vasodilation) for a given circulating blood volume (e.g., heat stress, hypoxia, sepsis). This paper examines the physiology of hypovolemia and its association with health and performance problems common to occupational, military and sports medicine. We discuss the maturation of individual-specific compensatory reserve or decompensation measures for future wearable sensor systems to effectively manage these hypovolemia problems. The paper then presents areas of future work to allow such technologies to translate from lab settings to use as decision aids for managing hypovolemia. We envision a future that incorporates elements of the compensatory reserve measure with advances in sensing technology and multiple modalities of cardiovascular sensing, additional contextual measures, and advanced noise reduction algorithms into a fully wearable system, creating a robust and physiologically sound approach to manage physical work, fatigue, safety and health issues associated with hypovolemia for workers, warfighters and athletes in austere conditions.

This paper proposes a new technique for performing 3D static-point cloud registration after calibrating a multi-view RGB-D camera using a 3D (dimensional) joint set. Consistent feature points are required to calibrate a multi-view camera, and accurate feature points are necessary to obtain high-accuracy calibration results. In general, a special tool, such as a chessboard, is used to calibrate a multi-view camera. However, this paper uses joints on a human skeleton as feature points for calibrating a multi-view camera to perform calibration efficiently without special tools. We propose an RGB-D-based calibration algorithm that uses the joint coordinates of the 3D joint set obtained through pose estimation as feature points. Since human body information captured by the multi-view camera may be incomplete, a joint set predicted based on image information obtained through this may be incomplete. After efficiently integrating a plurality of incomplete joint sets into one joint set, multi-view cameras can be calibrated by using the combined joint set to obtain extrinsic matrices. To increase the accuracy of calibration, multiple joint sets are used for optimization through temporal iteration. We prove through experiments that it is possible to calibrate a multi-view camera using a large number of incomplete joint sets.

The bearing is an essential component of a rotating machine. Sudden failure of the bearing may cause an unwanted breakdown of the manufacturing plant. In this paper, an intelligent fault diagnosis technique was developed to diagnose various faults that occur in a deep groove ball bearing. An experimental setup was designed and developed to generate faulty data in various conditions, such as inner race fault, outer race fault, and cage fault, along with the healthy condition. The time waveform of raw vibration data generated from the system was transformed into a frequency spectrum using the fast Fourier transform (FFT) method. These FFT signals were analyzed to detect the defective bearing. Another significant contribution of this paper is the application of a machine learning (ML) algorithm to diagnose bearing faults. The support vector machine (SVM) was used as the primary algorithm. As the efficiency of SVM heavily depends on hyperparameter tuning and optimum feature selection, the particle swarm optimization (PSO) technique was used to improve the model performance. The classification accuracy obtained using SVM with a traditional grid search cross-validation (CV) optimizer was 92%, whereas the improved accuracy using the PSO-based SVM was found to be 93.9%. The developed model was also compared with other traditional ML techniques such as k-nearest neighbor (KNN), decision tree (DT), and linear discriminant analysis (LDA). In every case, the proposed model outperformed the existing algorithms.

This paper considers the use of a machine learning system for the reconstruction and recognition of distorted or damaged patterns, in particular, images of faces partially covered with masks. The most up-to-date image reconstruction structures are based on constrained optimization algorithms and suitable regularizers. In contrast with the above-mentioned image processing methods, the machine learning system presented in this paper employs the superposition of system vectors setting up asymptotic centers of attraction. The structure of the system is implemented using Hopfield-type neural network-based biorthogonal transformations. The reconstruction property gives rise to a superposition processor and reversible computations. Moreover, this paper’s distorted image reconstruction sets up associative memories where images stored in memory are retrieved by distorted/inpainted key images.

The frequency estimation of complex exponential carrier signals in noise is a critical problem in signal processing. To solve this problem, a new iterative frequency estimator is presented in this paper. By iteratively computing the interpolation of DTFT samples, the proposed algorithm obtains a fine frequency estimate. In addition, its mean square error (MSE) analysis is presented in this paper. By analyzing influences of the selectable parameters on the estimation accuracy of the model, a method for choosing appropriate parameters is discussed, helping to reduce the estimation error of the proposed estimator. Simulation results show that compared with other algorithms with a comparable estimation accuracy, the proposed iterative estimator can obtain a root mean square error (RMSE) that is closer to Cramér–Rao lower bound (CRLB).

Neural networks are popular and useful in many fields, but they have the problem of giving high confidence responses for examples that are away from the training data. This makes the neural networks very confident in their prediction while making gross mistakes, thus limiting their reliability for safety-critical applications such as autonomous driving, space exploration, etc. This paper introduces a novel neuron generalization that has the standard dot-product-based neuron and the {\color{black} radial basis function (RBF)} neuron as two extreme cases of a shape parameter. Using a rectified linear unit (ReLU) as the activation function results in a novel neuron that has compact support, which means its output is zero outside a bounded domain. To address the difficulties in training the proposed neural network, it introduces a novel training method that takes a pretrained standard neural network that is fine-tuned while gradually increasing the shape parameter to the desired value. The theoretical findings of the paper are a bound on the gradient of the proposed neuron and a proof that a neural network with such neurons has the universal approximation property. This means that the network can approximate any continuous and integrable function with an arbitrary degree of accuracy. The experimental findings on standard benchmark datasets show that the proposed approach has smaller test errors than state-of-the-art competing methods and outperforms the competing methods in detecting out-of-distribution samples on two out of three datasets., Comment: 13 pages, 6 figures

Light-weight and accurate mapping is made possible by high-level feature extraction from sensor readings. In this paper, the high-level B-spline features from a 2D LIDAR are extracted with a faster method as a solution to the mapping problem, making it possible for the robot to interact with its environment while navigating. The computation time of feature extraction is very crucial when mobile robots perform real-time tasks. In addition to the existing assessment measures of B-spline feature extraction methods, the paper also includes a new benchmark time metric for evaluating how well the extracted features perform. For point-to-point association, the most reliable vertex control points of the spline features generated from the hints of low-level point feature FALKO were chosen. The standard three indoor and one outdoor data sets were used for the experiment. The experimental results based on benchmark performance metrics, specifically computation time, show that the presented approach achieves better results than the state-of-the-art methods for extracting B-spline features. The classification of the methods implemented in the B-spline features detection and the algorithms are also presented in the paper.

Recent methods for automatic blood vessel segmentation from fundus images have been commonly implemented as convolutional neural networks. While these networks report high values for objective metrics, the clinical viability of recovered segmentation masks remains unexplored. In this paper, we perform a pilot study to assess the clinical viability of automatically generated segmentation masks in the diagnosis of diseases affecting retinal vascularization. Five ophthalmologists with clinical experience were asked to participate in the study. The results demonstrate low classification accuracy, inferring that generated segmentation masks cannot be used as a standalone resource in general clinical practice. The results also hint at possible clinical infeasibility in experimental design. In the follow-up experiment, we evaluate the clinical quality of masks by having ophthalmologists rank generation methods. The ranking is established with high intra-observer consistency, indicating better subjective performance for a subset of tested networks. The study also demonstrates that objective metrics are not correlated with subjective metrics in retinal segmentation tasks for the methods involved, suggesting that objective metrics commonly used in scientific papers to measure the method’s performance are not plausible criteria for choosing clinically robust solutions.

Airborne radars are susceptible to a large number of clutter, noise and variable jamming signals in the real environment, especially when faced with active main lobe jamming, as the waveform shortcut technology in the traditional regime can no longer meet the actual battlefield radar anti-jamming requirements. Therefore, it is necessary to study anti-main-lobe jamming techniques for airborne radars in complex environments to improve their battlefield survivability. In this paper, we propose an airborne radar waveform design method based on a deep reinforcement learning (DRL) algorithm under clutter and jamming conditions, after previous research on reinforcement-learning (RL)-based airborne radar anti-jamming waveform design methods that have improved the anti-jamming performance of airborne radars. The method uses a Markov decision process (MDP) to describe the complex operating environment of airborne radars, calculates the value of the radar anti-jamming waveform strategy under various jamming states using deep neural networks and designs the optimal anti-jamming waveform strategy for airborne radars based on the duelling double deep Q network (D3QN) algorithm. In addition, the method uses an iterative transformation method (ITM) to generate the time domain signals of the optimal waveform strategy. Simulation results show that the airborne radar waveform designed based on the deep reinforcement learning algorithm proposed in this paper improves the signal-to-jamming plus noise ratio (SJNR) by 2.08 dB and 3.03 dB, and target detection probability by 26.79% and 44.25%, respectively, compared with the waveform designed based on the reinforcement learning algorithm and the conventional linear frequency modulation (LFM) signal at a radar transmit power of 5 W. The airborne radar waveform design method proposed in this paper helps airborne radars to enhance anti-jamming performance in complex environments while further improving target detection performance.

Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). This paper presents the design of a safety system based on controlled electromagnetic field (CEMF) sensing technology to prevent accidents caused by power tools, especially related to handheld circular saws. The safety system creates an invisible protection bubble of electromagnetic field around the cutting edge. The system can provide early warning or critical warning when a person penetrates the safety bubble. This paper covers how the CEMF technology has been adapted to add value within this application where it needs to coexist with a difficult environment of metallic parts turning thousands of times per minute, strong vibrations, and different ranges of materials to be processed. The proposed contactless solution successfully detects the user, providing enough time for the power tool to totally stop its movement before touching and harming the user. This key property has required a careful optimization of the electromagnetic field generation, the design of a shield circuitry capable of operating properly in a large metal device, and the development of a multi-frame algorithm to address the stringent requirements related to the ability of the system to react to both very fast and very slow events. The feasibility of the system has been validated by a virtual testbench.

Traditional contact voltage measurement requires a direct electrical connection to the system, which is not easy to install and maintain. The voltage measurement based on the electric field coupling plate capacitance structure does not need to be in contact with the measured object or the ground, which can avoid the above problems. However, most of the existing flat-plate structure voltage measurement sensors are not only expensive to manufacture, but also bulky, and when the relative position between the wire under test and the sensor changes, it will bring great measurement errors, making it difficult to meet actual needs. Aiming to address the above problems, this paper proposes a multi-electrode array structure non-contact voltage sensor and signal processing algorithm. The sensor is manufactured by the PCB process, which effectively reduces the manufacturing cost and process difficulty. The experimental and simulation results show that, when the relative position of the wire and the sensor is offset by 10 mm in the 45° direction, the relative error of the traditional single-electrode voltage sensor is 17.62%, while the relative error of the multi-electrode voltage sensor designed in this paper is only 0.38%. In addition, the ratio error of the sensor under the condition of power frequency of 50 Hz is less than ±1% and the phase difference is less than 4°. The experimental results show that the sensor has good accuracy and linearity.

Tool wear condition monitoring during the machining process is one of the most important considerations in precision manufacturing. Cutting force is one of the signals that has been widely used for tool wear condition monitoring, which contains the dynamical information of tool wear conditions. This paper proposes a novel multivariate cutting force-based tool wear monitoring method using one-dimensional convolutional neural network (1D CNN). Firstly, multivariate variational mode decomposition (MVMD) is used to process the multivariate cutting force signals. The multivariate band-limited intrinsic mode functions (BLIMFs) are obtained, which contain a large number of nonlinear and nonstationary tool wear characteristics. Afterwards, the proposed modified multiscale permutation entropy (MMPE) is used to measure the complexity of multivariate BLIMFs. The entropy values on multiple scales are calculated as condition indicators in tool wear condition monitoring. Finally, the one-dimensional feature vectors are constructed and employed as the input of 1D CNN to achieve accurate and stable tool wear condition monitoring. The results of the research in this paper demonstrate that the proposed approach has broad prospects in tool wear condition monitoring.

Over the last few decades, computer-aided diagnosis systems have become a part of clinical practice. They have the potential to assist clinicians in daily diagnostic tasks. The image processing techniques are fast, repeatable, and robust, which helps physicians to detect, classify, segment, and measure various structures. The recent rapid development of computer methods for high-frequency ultrasound image analysis opens up new diagnostic paths in dermatology, allergology, cosmetology, and aesthetic medicine. This paper, being the first in this area, presents a research overview of high-frequency ultrasound image processing techniques, which have the potential to be a part of computer-aided diagnosis systems. The reviewed methods are categorized concerning the application, utilized ultrasound device, and image data-processing type. We present the bridge between diagnostic needs and already developed solutions and discuss their limitations and future directions in high-frequency ultrasound image analysis. A search was conducted of the technical literature from 2005 to September 2022, and in total, 31 studies describing image processing methods were reviewed. The quantitative and qualitative analysis included 39 algorithms, which were selected as the most effective in this field. They were completed by 20 medical papers and define the needs and opportunities for high-frequency ultrasound application and CAD development.

Ride-hailed shared autonomous vehicles (SAV) have emerged recently as an economically feasible way of introducing autonomous driving technologies while serving the mobility needs of under-served communities. There has also been corresponding research work on optimization of the operation of these SAVs. However, the current state-of-the-art research in this area treats very simple networks, neglecting the effect of a realistic other traffic representation, and is not useful for planning deployments of SAV service. In contrast, this paper utilizes a recent autonomous shuttle deployment site in Columbus, Ohio, as a basis for mobility studies and the optimization of SAV fleet deployment. Furthermore, this paper creates an SAV dispatcher based on reinforcement learning (RL) to minimize passenger wait time and to maximize the number of passengers served. The created taxi-dispatcher is then simulated in a realistic scenario while avoiding generalization or over-fitting to the area. It is found that an RL-aided taxi dispatcher algorithm can greatly improve the performance of a deployment of SAVs by increasing the overall number of trips completed and passengers served while decreasing the wait time for passengers.

Sugarcane stem node identification is the core technology required for the intelligence and mechanization of the sugarcane industry. However, detecting stem nodes quickly and accurately is still a significant challenge. In this paper, in order to solve this problem, a new algorithm combining YOLOv3 and traditional methods of computer vision is proposed, which can improve the identification rate during automated cutting. First, the input image is preprocessed, during which affine transformation is used to correct the posture of the sugarcane and a rotation matrix is established to obtain the region of interest of the sugarcane. Then, a dataset is built to train the YOLOv3 network model and the position of the stem nodes is initially determined using the YOLOv3 model. Finally, the position of the stem nodes is further located accurately. In this step, a new gradient operator is proposed to extract the edge of the image after YOLOv3 recognition. Then, a local threshold determination method is proposed, which is used to binarize the image after edge extraction. Finally, a localization algorithm for stem nodes is designed to accurately determine the number and location of the stem nodes. The experimental results show that the precision rate, recall rate, and harmonic mean of the stem node recognition algorithm in this paper are 99.68%, 100%, and 99.84%, respectively. Compared to the YOLOv3 network, the precision rate and the harmonic mean are improved by 2.28% and 1.13%, respectively. Compared to other methods introduced in this paper, this algorithm has the highest recognition rate.

Optical transport networks (OTNs) are widely used in backbone- and metro-area transmission networks to increase network transmission capacity. In the OTN, it is particularly crucial to rationally allocate routes and maximize network capacities. By employing deep reinforcement learning (DRL)- and software-defined networking (SDN)-based solutions, the capacity of optical networks can be effectively increased. However, because most DRL-based routing optimization methods have low sample usage and difficulty in coping with sudden network connectivity changes, converging in software-defined OTN scenarios is challenging. Additionally, the generalization ability of these methods is weak. This paper proposes an ensembles- and message-passing neural-network-based Deep Q-Network (EMDQN) method for optical network routing optimization to address this problem. To effectively explore the environment and improve agent performance, the multiple EMDQN agents select actions based on the highest upper-confidence bounds. Furthermore, the EMDQN agent captures the network’s spatial feature information using a message passing neural network (MPNN)-based DRL policy network, which enables the DRL agent to have generalization capability. The experimental results show that the EMDQN algorithm proposed in this paper performs better in terms of convergence. EMDQN effectively improves the throughput rate and link utilization of optical networks and has better generalization capabilities.

Background: Digital clinical measures collected via various digital sensing technologies such as smartphones, smartwatches, wearables, ingestible, and implantable are increasingly used by individuals and clinicians to capture health outcomes or behavioral and physiological characteristics of individuals. Time series classification (TSC) is very commonly used in modeling digital clinical measures. While deep learning models for TSC are very common and powerful, there exist some fundamental challenges. This review presents non-deep learning models commonly used for time series classification in biomedical applications that achieve high performance. Objective: We performed a systematic review to characterize the techniques used in time series classification of digital clinical measures throughout all stages of data processing and model building. Methods: We conducted a literature search on PubMed, and the Institute of Electrical and Electronics Engineers (IEEE), Web of Science, and SCOPUS databases using a range of search terms to retrieve peer-reviewed articles reporting academic research on digital clinical measures in the five year period between June 2016 and June 2021. We identified and categorized research studies based on the types of classification algorithms and sensor input types. Results: We found 452 papers in total from four different databases: PubMed, IEEE, Web of Science Database, and SCOPUS. After removing duplicates and irrelevant papers, 135 articles remained for detailed review and data extraction. Among these, engineered features using time series methods that were subsequently fed into widely-used machine learning classifiers was the most commonly used technique and also most frequently achieved the best performance metrics (77 out of 135 articles). Statistical modeling (24 out of 135 articles) algorithms were the second most common and also second best classification technique. Wavelet-based classification models (8 out of 135 articles) were also common. Electroencephalogram (29 out of 135 articles) was the most common data type used as an input. Accuracy was the most commonly reported performance metric, with 67.65% of articles reporting on accuracy. In this review paper, we provide summaries of signal pre-processing methods, feature engineering and selection methods, time series models, as well as model interpretations. Importantly, we found that about 50% of the papers only report one performance metric, which may result in a skewed view of overall performance. Conclusion: While high time series classification performance has been achieved in digital clinical, physiological, or biomedical measures, no standard benchmark datasets, modeling methods, or reporting methodology exist. There is no single widely used method for time series model development or feature interpretation– many different methods have proven successful.

Motion blur recovery is a common method in the field of remote sensing image processing that can effectively improve the accuracy of detection and recognition. Among the existing motion blur recovery methods, the algorithms based on deep learning do not rely on a priori knowledge and, thus, have better generalizability. However, the existing deep learning algorithms usually suffer from feature misalignment, resulting in a high probability of missing details or errors in the recovered images. This paper proposes an end-to-end generative adversarial network (SDD-GAN) for single-image motion deblurring to address this problem and to optimize the recovery of blurred remote sensing images. Firstly, this paper applies a feature alignment module (FAFM) in the generator to learn the offset between feature maps to adjust the position of each sample in the convolution kernel and to align the feature maps according to the context; secondly, a feature importance selection module is introduced in the generator to adaptively filter the feature maps in the spatial and channel domains, preserving reliable details in the feature maps and improving the performance of the algorithm. In addition, this paper constructs a self-constructed remote sensing dataset (RSDATA) based on the mechanism of image blurring caused by the high-speed orbital motion of satellites. Comparative experiments are conducted on self-built remote sensing datasets and public datasets as well as on real remote sensing blurred images taken by an in-orbit satellite (CX-6(02)). The results show that the algorithm in this paper outperforms the comparison algorithm in terms of both quantitative evaluation and visual effects.

This paper presents a method to include unmodeled dynamics of load or a robot’s end-effector into algorithms for collision detection or general understanding of a robot’s operation context. The approach relies on the application of a previously developed modification of the Dynamic Time Warping algorithm, as well as a universally applicable algorithm for identifying kinematic parameters. The entire process can be applied to arbitrary robot configuration, and it does not require identification of dynamic parameters. The paper addresses the two main categories of contact tasks with unmodelled dynamics, which are determined based on whether the external contact force has a consistent profile in the end effector or base coordinate. Conclusions for representative examples analysed in the paper are applicable to tasks such as load manipulation, press bending, and crimping for the first type of forces and applications such as drilling, screwdriving, snap-fit, bolting, and riveting assembly for the latter category. The results presented in the paper are based on realistic testing with measurements obtained from an industrial robot.

While there are numerous causes of waste in the healthcare system, some of this waste is associated with inefficiency. Among the proposed solutions to address inefficiency is clinic layout optimization. Such optimization depends on how operating resources and instruments are placed in the clinic, in what order they are accessed to attain a particular task, and the mobility of clinicians between different clinic rooms to accomplish different clinic tasks. Traditionally, such optimization research involves manual monitoring by human proctors, which is time consuming, erroneous, unproductive, and subjective. If mobility patterns in an indoor space can be determined automatically in real time, layout and operation-related optimization decisions based on these patterns can be implemented accurately and continuously in a timely fashion. This paper explores this application domain where precise localization is not required; however, the determination of mobility is essential on a real-time basis. Given that, this research explores how only mobile devices and their built-in Bluetooth received signal strength indicator (RSSI) can be used to determine such mobility. With a collection of stationary mobile devices, with their computational and networking capabilities and lack of energy requirements, the mobility of moving mobile devices was determined. The research methodology involves developing two new algorithms that use raw RSSI data to create visualizations of movements across different operational units identified by stationary nodes. Compared with similar approaches, this research showcases that the method presented in this paper is viable and can produce mobility patterns in indoor spaces that can be utilized further for data analysis and visualization.

DNA (Deoxyribonucleic Acid) Cryptography has revolutionized information security by combining rigorous biological and mathematical concepts to encode original information in terms of a DNA sequence. Such schemes are crucially dependent on corresponding DNA-based cryptographic keys. However, owing to the redundancy or observable patterns, some of the keys are rendered weak as they are prone to intrusions. This paper proposes a Genetic Algorithm inspired method to strengthen weak keys obtained from Random DNA-based Key Generators instead of completely discarding them. Fitness functions and the application of genetic operators have been chosen and modified to suit DNA cryptography fundamentals in contrast to fitness functions for traditional cryptographic schemes. The crossover and mutation rates are reducing with each new population as more keys are passing fitness tests and need not be strengthened. Moreover, with the increasing size of the initial key population, the key space is getting highly exhaustive and less prone to Brute Force attacks. The paper demonstrates that out of an initial 25 × 25 population of DNA Keys, 14 keys are rendered weak. Complete results and calculations of how each weak key can be strengthened by generating 4 new populations are illustrated. The analysis of the proposed scheme for different initial populations shows that a maximum of 8 new populations has to be generated to strengthen all 500 weak keys of a 500 × 500 initial population.

The paper presents research on a specific approach to the issue of computed tomography with an incomplete data set. The case of incomplete information is quite common, for example when examining objects of large size or difficult to access. Algorithms devoted to this type of problems can be used to detect anomalies in coal seams that pose a threat to the life of miners. The most dangerous example of such an anomaly may be a compressed gas tank, which expands rapidly during exploitation, at the same time ejecting rock fragments, which are a real threat to the working crew. The approach presented in the paper is an improvement of the previous idea, in which the detected objects were represented by sequences of points. These points represent rectangles, which were characterized by sequences of their parameters. This time, instead of sequences in the representation, there are sets of objects, which allow for the elimination of duplicates. As a result, the reconstruction is faster. The algorithm presented in the paper solves the inverse problem of finding the minimum of the objective function. Heuristic algorithms are suitable for solving this type of tasks. The following heuristic algorithms are described, tested and compared: Aquila Optimizer (AQ), Firefly Algorithm (FA), Whale Optimization Algorithm (WOA), Butterfly Optimization Algorithm (BOA) and Dynamic Butterfly Optimization Algorithm (DBOA). The research showed that the best algorithm for this type of problem turned out to be DBOA.

Thermal inkjet printing can generate more than 300,000 droplets of picoliter scale within one second stably, and the image analysis workflow is used to quantify the positive and negative values of the droplets. In this paper, the SimpleBlobDetector detection algorithm is used to identify and localize droplets with a volume of 24 pL in bright field images and suppress bright spots and scratches when performing droplet location identification. The polynomial surface fitting of the pixel grayscale value of the fluorescence channel image can effectively compensate and correct the image vignetting caused by the optical path, and the compensated fluorescence image can accurately classify positive and negative droplets by the k-means clustering algorithm. 20 µL of the sample solution in the result reading chip can produce more than 100,000 effective droplets. The effective droplet identification correct rate of 20 images of random statistical samples can reach more than 99% and the classification accuracy of positive and negative droplets can reach more than 98% on average. This paper overcomes the problem of effectively classifying positive and negative droplets caused by the poor image quality of photographed picolitre ddPCR droplets caused by optical hardware limitations.

Aiming at the common problems, such as noise pollution, low contrast, and color distortion in underwater images, and the characteristics of holothurian recognition, such as morphological ambiguity, high similarity with the background, and coexistence of special ecological scenes, this paper proposes an underwater holothurian target-detection algorithm (FA-CenterNet), based on improved CenterNet and scene feature fusion. First, to reduce the model’s occupancy of embedded device resources, we use EfficientNet-B3 as the backbone network to reduce the model’s Params and FLOPs. At the same time, EfficientNet-B3 increases the depth and width of the model, which improves the accuracy of the model. Then, we design an effective FPT (feature pyramid transformer) combination module to fully focus and mine the information on holothurian ecological scenarios of different scales and spaces (e.g., holothurian spines, reefs, and waterweeds are often present in the same scenario as holothurians). The co-existing scene information can be used as auxiliary features to detect holothurians, which can improve the detection ability of fuzzy and small-sized holothurians. Finally, we add the AFF module to realize the deep fusion of the shallow-detail and high-level semantic features of holothurians. The results show that the method presented in this paper yields better results on the 2020 CURPC underwater target-detection image dataset with an AP50 of 83.43%, Params of 15.90 M, and FLOPs of 25.12 G compared to other methods. In the underwater holothurian-detection task, this method improves the accuracy of detecting holothurians with fuzzy features, a small size, and dense scene. It also achieves a good balance between detection accuracy, Params, and FLOPs, and is suitable for underwater holothurian detection in most situations.

The combination of unmanned aerial vehicles (UAVs) and artificial intelligence is significant and is a key topic in recent substation inspection applications; and meter reading is one of the challenging tasks. This paper proposes a method based on the combination of YOLOv5s object detection and Deeplabv3+ image segmentation to obtain meter readings through the post-processing of segmented images. Firstly, YOLOv5s was introduced to detect the meter dial area and the meter was classified. Following this, the detected and classified images were passed to the image segmentation algorithm. The backbone network of the Deeplabv3+ algorithm was improved by using the MobileNetv2 network, and the model size was reduced on the premise that the effective extraction of tick marks and pointers was ensured. To account for the inaccurate reading of the meter, the divided pointer and scale area were corroded first, and then the concentric circle sampling method was used to flatten the circular dial area into a rectangular area. Several analog meter readings were calculated by flattening the area scale distance. The experimental results show that the mean average precision of 50 (mAP50) of the YOLOv5s model with this method in this data set reached 99.58%, that the single detection speed reached 22.2 ms, and that the mean intersection over union (mIoU) of the image segmentation model reached 78.92%, 76.15%, 79.12%, 81.17%, and 75.73%, respectively. The single segmentation speed reached 35.1 ms. At the same time, the effects of various commonly used detection and segmentation algorithms on the recognition of meter readings were compared. The results show that the method in this paper significantly improved the accuracy and practicability of substation meter reading detection in complex situations.

Existing wearable systems that use G-sensors to identify daily activities have been widely applied for medical, sports and military applications, while body temperature as an obvious physical characteristic that has rarely been considered in the system design and relative applications of HAR. In the context of the normalization of COVID-19, the prevention and control of the epidemic has become a top priority. Temperature monitoring plays an important role in the preliminary screening of the population for fever. Therefore, this paper proposes a wearable device embedded with inertial and temperature sensors that is used to apply human behavior recognition (HAR) to body surface temperature detection for body temperature monitoring and adjustment by evaluating recognition algorithms. The sensing system consists of an STM 32-based microcontroller, a 6-axis (accelerometer and gyroscope) sensor, and a temperature sensor to capture the original data from 10 individual participants under 4 different daily activity scenarios. Then, the collected raw data are pre-processed by signal standardization, data stacking and resampling. For HAR, several machine learning (ML) and deep learning (DL) algorithms are implemented to classify the activities. To compare the performance of different classifiers on the seven-dimensional dataset with temperature sensing signals, evaluation metrics and the algorithm running time are considered, and random forest (RF) is found to be the best-performing classifier with 88.78% recognition accuracy, which is higher than the case of the absence of temperature data (&lt, 78%). In addition, the experimental results show that participants’ body surface temperature in dynamic activities was lower compared to sitting, which can be associated with the possible missing fever population due to temperature deviations in COVID-19 prevention. According to different individual activities, epidemic prevention workers are supposed to infer the corresponding standard normal body temperature of a patient by referring to the specific values of the mean expectation and variance in the normal distribution curve provided in this paper.

The safety and reliability of the power grid are related to national power security, economic development and people’s daily life. The occurrence of extreme weather changes the external environment greatly. Including generators and transmission lines, many power grid units cannot resist such a huge attack and get damaged easily, which forces units to quit from the power grid running system for a while. Furthermore, if the number of influenced units is high enough, the whole power system will be destroyed by cascading failure caused by extreme weather. Aiming at dealing with the cascading failure emergencies, this paper is trying to improve the traditional power structural vulnerability model so that it can be used to discuss extreme weather and propose a theoretical topological model to help scholars measure the damage caused by extreme cases. Based on previous research in this field, this paper utilizes complex network knowledge to build the power grid topology model. Then, considering extreme cases and the three attack modes simulation process, this paper makes use of the characteristic parameters of the power grid topology model and designs an algorithm, according to the realistic situation of the propagation mechanism of cascading failure of the power grid model as well as extreme weather research. Finally, taking IEEE-30 and IEEE-118 node bus system as examples, which shows that the structural vulnerability method proposed in this paper can properly address the mechanism of unbalanced load of cascading failure of power grid units under extreme conditions and can provide theoretical reference for preventing and reducing the impact of extreme cases on power grid which improves the reliability of the power grid.

As is well-known, defects precisely affect the lives and functions of the machines in which they occur, and even cause potentially catastrophic casualties. Therefore, quality assessment before mounting is an indispensable requirement for factories. Apart from the recognition accuracy, current networks suffer from excessive computing complexity, making it of great difficulty to deploy in the manufacturing process. To address these issues, this paper introduces binary networks into the area of surface defect detection for the first time, for the reason that binary networks prohibitively constrain weight and activation to +1 and −1. The proposed Bi-ShuffleNet and U-BiNet utilize binary convolution layers and activations in low bitwidth, in order to reach comparable performances while incurring much less computational cost. Extensive experiments are conducted on real-life NEU and Magnetic Tile datasets, revealing the least OPs required and little accuracy decline. When classifying the defects, Bi-ShuffleNet yields comparable results to counterpart networks, with at least 2× inference complexity reduction. Defect segmentation results indicate similar observations. Some network design rules in defect detection and binary networks are also summarized in this paper.

As a typical application of sensor networks, there exist many information security problems in smart parks, such as confusion of personnel access, lack of security management, disorderly data flow, insufficient collection of audit evidence, and so on. Aiming at the scenario of personnel and equipment moving in different areas of smart parks, the paper proposes a joint authorization and dynamic access control mechanism, which can provide unified identity management services, access control services, and policy management services, and effectively solve the problem of multi-authorization in user identity and authority management. The license negotiation interaction protocol is designed to prevent common network attack threats in the process of identity authentication and authority management. In order to realize the tamper-proof storage of personnel and equipment movement trajectory, the paper also designs a movement trajectory traceability protocol based on a Merkle tree, which solves the problems of internal personnel malicious attack, trusted third-party dependency bottleneck, high overheads of tracking algorithms, and so on. The experimental results show that compared with the current security control mechanisms for sensor networks, the joint authorization, and dynamic access control mechanism can support multi-party authorization and traceability, while the overhead it generates in initialization, encryption, decryption, and key generation steps are basically the same as other mechanisms do.

Autonomous Vehicles (AVs) have the potential to solve many traffic problems, such as accidents, congestion and pollution. However, there are still challenges to overcome, for instance, AVs need to accurately perceive their environment to safely navigate in busy urban scenarios. The aim of this paper is to review recent articles on computer vision techniques that can be used to build an AV perception system. AV perception systems need to accurately detect non-static objects and predict their behaviour, as well as to detect static objects and recognise the information they are providing. This paper, in particular, focuses on the computer vision techniques used to detect pedestrians and vehicles. There have been many papers and reviews on pedestrians and vehicles detection so far. However, most of the past papers only reviewed pedestrian or vehicle detection separately. This review aims to present an overview of the AV systems in general, and then review and investigate several detection computer vision techniques for pedestrians and vehicles. The review concludes that both traditional and Deep Learning (DL) techniques have been used for pedestrian and vehicle detection; however, DL techniques have shown the best results. Although good detection results have been achieved for pedestrians and vehicles, the current algorithms still struggle to detect small, occluded, and truncated objects. In addition, there is limited research on how to improve detection performance in difficult light and weather conditions. Most of the algorithms have been tested on well-recognised datasets such as Caltech and KITTI; however, these datasets have their own limitations. Therefore, this paper recommends that future works should be implemented on more new challenging datasets, such as PIE and BDD100K. EPSRC DTP PhD studentship

In this paper a modified wavelet synthesis algorithm for continuous wavelet transform is proposed, allowing one to obtain a guaranteed approximation of the maternal wavelet to the sample of the analyzed signal (overlap match) and, at the same time, a formalized representation of the wavelet. What distinguishes this method from similar ones? During the procedure of wavelets’ synthesis for continuous wavelet transform it is proposed to use splines and artificial neural networks. The paper also suggests a comparative analysis of polynomial, neural network, and wavelet spline models. It also deals with feasibility of using these models in the synthesis of wavelets during such studies like fine structure of signals, as well as in analysis of large parts of signals whose shape is variable. A number of studies have shown that during the wavelets’ synthesis, the use of artificial neural networks (based on radial basis functions) and cubic splines enables the possibility of obtaining guaranteed accuracy in approaching the maternal wavelet to the signal’s sample (with no approximation error). It also allows for its formalized representation, which is especially important during software implementation of the algorithm for calculating the continuous conversions at digital signal processors and microcontrollers. This paper demonstrates the possibility of using synthesized wavelet, obtained based on polynomial, neural network, and spline models, during the performance of an inverse continuous wavelet transform.

The proposed StegoFrameOrder (SFO) method enables the transmission of covert data in wireless computer networks exploiting non-deterministic algorithms of medium access (such as the distributed coordination function), especially in IEEE 802.11 networks. Such a covert channel enables the possibility of leaking crucial information outside secured network in a manner that is difficult to detect. The SFO method embeds hidden bits of information in the relative order of frames transmitted by wireless terminals operating on the same radio channel. The paper presents an idea of this covert channel, its implementation, and possible variants. The paper also discusses implementing the SFO method in a real environment and the experiments performed in the real-world scenario.

In this paper, comprehensive mPoint, a method for generating 3D (range, azimuth, and elevation) point cloud of human targets using a Frequency-Modulated Continuous Wave (FMCW) signal and Multi-Input Multi-Output (MIMO) millimeter wave radar is proposed. Distinct from the TI-mPoint method proposed by TI technology, a comprehensive mPoint method considering both the static and dynamic characteristics of radar reflected signals is utilized to generate a high precision point cloud, resulting in more comprehensive information of the target being detected. The radar possessing 60&ndash, 64 GHz FMCW signal with two sets of different dimensional antennas is utilized in order to experimentally verify the results of the methodology. By using the proposed process, the point cloud data of human targets can be obtained based on six different postures of the underlying human body. The human posture cube and point cloud accuracy rates are defined in the paper in order to quantitively and qualitatively evaluate the quality of the generated point cloud. Benefitting from the proposed comprehensive mPoint, evidence shows that the point number and the accuracy rate of the generated point cloud compared with those from the popular TI-mPoint can be largely increased by 86% and 42%, respectively. In addition, the noise level of multipath reflection can be effectively reduced. Moreover, the length of the algorithm running time is only 1.6% longer than that of the previous method as a slight tradeoff.

The increasing number of Android malware forced antivirus (AV) companies to rely on automated classification techniques to determine the family and class of suspicious samples. The research community relies heavily on such labels to carry out prevalence studies of the threat ecosystem and to build datasets that are used to validate and benchmark novel detection and classification methods. In this work, we carry out an extensive study of the Android malware ecosystem by surveying white papers and reports from 6 key players in the industry, as well as 81 papers from 8 top security conferences, to understand how malware datasets are used by both. We, then, explore the limitations associated with the use of available malware classification services, namely VirusTotal (VT) engines, for determining the family of an Android sample. Using a dataset of 2.47 M Android malware samples, we find that the detection coverage of VT's AVs is generally very low, that the percentage of samples flagged by any 2 AV engines does not go beyond 52%, and that common families between any pair of AV engines is at best 29%. We rely on clustering to determine the extent to which different AV engine pairs agree upon which samples belong to the same family (regardless of the actual family name) and find that there are discrepancies that can introduce noise in automatic label unification schemes. We also observe the usage of generic labels and inconsistencies within the labels of top AV engines, suggesting that their efforts are directed towards accurate detection rather than classification. Our results contribute to a better understanding of the limitations of using Android malware family labels as supplied by common AV engines. This work has been supported by the “Ramon y Cajal” Fellowship RYC-2020-029401.

This paper introduces an ambient light rejection (ALR) circuit for the autonomous adaptation of a subretinal implant system. The sub-retinal implants, located beneath a bipolar cell layer, are known to have a significant advantage in spatial resolution by integrating more than a thousand pixels, compared to epi-retinal implants. However, challenges remain regarding current dispersion in high-density retinal implants, and ambient light induces pixel saturation. Thus, the technical issues of ambient light associated with a conventional image processing technique, which lead to high power consumption and area occupation, are still unresolved. Thus, it is necessary to develop a novel image-processing unit to handle ambient light, considering constraints related to power and area. In this paper, we present an ALR circuit as an image-processing unit for sub-retinal implants. We first introduced an ALR algorithm to reduce the ambient light in conventional retinal implants, next, we implemented the ALR algorithm as an application-specific integrated chip (ASIC). The ALR circuit was fabricated using a standard 0.35-μm CMOS process along with an image-sensor-based stimulator, a sensor pixel, and digital blocks. As experimental results, the ALR circuit occupies an area of 190 µm2, consumes a power of 3.2 mW and shows a maximum response time of 1.6 s at a light intensity of 20,000 lux. The proposed ALR circuit also has a pixel loss rate of 0.3%. The experimental results show that the ALR circuit leads to a sensor pixel (SP) being autonomously adjusted, depending on the light intensity.

This paper presents a new approach for the optimization of a dictionary used in ECG signal compression and reconstruction systems, based on Compressed Sensing (CS). Alternatively to fully data driven methods, which learn the dictionary from the training data, the proposed approach uses an over complete wavelet dictionary, which is then reduced by means of a training phase. Moreover, the alignment of the frames according to the position of the R-peak is proposed, such that the dictionary optimization can exploit the different scaling features of the ECG waves. Therefore, at first, a training phase is performed in order to optimize the overcomplete dictionary matrix by reducing its number of columns. Then, the optimized matrix is used in combination with a dynamic sensing matrix to compress and reconstruct the ECG waveform. In this paper, the mathematical formulation of the patient-specific optimization is presented and three optimization algorithms have been evaluated. For each of them, an experimental tuning of the convergence parameter is carried out, in order to ensure that the algorithm can work in its most suitable conditions. The performance of each considered algorithm is evaluated by assessing the Percentage of Root-mean-squared Difference (PRD) and compared with the state of the art techniques. The obtained experimental results demonstrate that: (i) the utilization of an optimized dictionary matrix allows a better performance to be reached in the reconstruction quality of the ECG signals when compared with other methods, (ii) the regularization parameters of the optimization algorithms should be properly tuned to achieve the best reconstruction results, and (iii) the Multiple Orthogonal Matching Pursuit (M-OMP) algorithm is the better suited algorithm among those examined.

Technology advances have made possible improvements such as Continuous Glucose Monitors, giving the patient a glucose reading every few minutes, or insulin pumps, allowing more personalized therapies. With the increasing number of available closed-loop systems, new challenges appear regarding algorithms and functionalities. Several of the analysed systems in this paper try to adapt to changes in some patients’ conditions and, in several of these systems, other variables such as basal needs are considered fixed from day to day to simplify the control problem. Therefore, these systems require a correct adjustment of the basal needs profile which becomes crucial to obtain good results. In this paper a novel approach tries to dynamically determine the insulin basal needs of the patient and use this information within a closed-loop algorithm, allowing the system to dynamically adjust in situations of illness, exercise, high-fat-content meals or even partially blocked infusion sites and avoiding the need for setting a basal profile that approximately matches the basal needs of the patient. The insulin sensitivity factor and the glycemic target are also dynamically modified according to the situation of the patient. Basal insulin needs are dynamically determined through linear regression via the decomposition of previously dosed insulin and its effect on the patient’s glycemia. Using the obtained value as basal insulin needs and other mechanisms such as basal needs modification through its trend, ISF and glycemic targets modification and low-glucose-suspend threshold, the safety of the algorithm is improved. The dynamic basal insulin needs determination was successfully included in a closed-loop control algorithm and was simulated on 30 virtual patients (10 adults, 10 adolescent and 10 children) using an open-source python implementation of the FDA-approved (Food and Drug Administration) UVa (University of Virginia)/Padova Simulator. Simulations showed that the proposed system dynamically determines the basal needs and can adapt to a partial blockage of the insulin infusion, obtaining similar results in terms of time in range to the case in which no blockage was simulated. The proposed algorithm can be incorporated to other current closed-loop control algorithms to directly estimate the patient’s basal insulin needs or as a monitoring channel to detect situations in which basal needs may differ from the expected ones.

In an autonomous vehicle, the lane following algorithm is an important component, which is a basic function of autonomous driving. However, the existing lane following system has a few shortcomings: first, the control method it adopts requires an accurate system model, and different vehicles have different parameters, which needs a lot of parameter calibration work. The second is that it may fail on road sections where the lateral acceleration requirements of vehicles are large, such as large curves. Third, its decision-making system is defined based on rules, which has disadvantages: it is difficult to formulate, human subjective factors cannot guarantee objectivity, coverage is difficult to guarantee. In recent years, the deep deterministic policy gradient (DDPG) algorithm has been widely used in the field of autonomous driving due to its strong nonlinear fitting ability and generalization performance. However, the DDPG algorithm has overestimated state action values and large cumulative errors, low training efficiency and other issues. Therefore, this paper improves the DDPG algorithm based on the double critic networks and priority experience replay mechanism. Then this paper proposes a lane following method based on this algorithm. Experiment shows that the algorithm can achieve excellent following results under various road conditions.

This paper addresses the problem of unsupervised video summarization. Video summarization helps people browse large-scale videos easily with a summary from the selected frames of the video. In this paper, we propose an unsupervised video summarization method with piecewise linear interpolation (Interp-SUM). Our method aims to improve summarization performance and generate a natural sequence of keyframes with predicting importance scores of each frame utilizing the interpolation method. To train the video summarization network, we exploit a reinforcement learning-based framework with an explicit reward function. We employ the objective function of the exploring under-appreciated reward method for training efficiently. In addition, we present a modified reconstruction loss to promote the representativeness of the summary. We evaluate the proposed method on two datasets, SumMe and TVSum. The experimental result showed that Interp-SUM generates the most natural sequence of summary frames than any other the state-of-the-art methods. In addition, Interp-SUM still showed comparable performance with the state-of-art research on unsupervised video summarization methods, which is shown and analyzed in the experiments of this paper.

Recent progress in deep learning has led to accurate and efficient generic object detection networks. Training of highly reliable models depends on large datasets with highly textured and rich images. However, in real-world scenarios, the performance of the generic object detection system decreases when (i) occlusions hide the objects, (ii) objects are present in low-light images, or (iii) they are merged with background information. In this paper, we refer to all these situations as challenging environments. With the recent rapid development in generic object detection algorithms, notable progress has been observed in the field of deep learning-based object detection in challenging environments. However, there is no consolidated reference to cover the state of the art in this domain. To the best of our knowledge, this paper presents the first comprehensive overview, covering recent approaches that have tackled the problem of object detection in challenging environments. Furthermore, we present a quantitative and qualitative performance analysis of these approaches and discuss the currently available challenging datasets. Moreover, this paper investigates the performance of current state-of-the-art generic object detection algorithms by benchmarking results on the three well-known challenging datasets. Finally, we highlight several current shortcomings and outline future directions.

This paper presents an automatic classification of plastic material’s inorganic pigment using terahertz spectroscopy and convolutional neural networks (CNN). The plastic materials were placed between the THz transmitter and receiver, and the acquired THz signals were classified using a supervised learning approach. A THz frequency band between 0.1–1.2 THz produced a one-dimensional (1D) vector that is almost impossible to classify directly using supervised learning. This paper proposes a novel pre-processing of 1D THz data that transforms 1D data into 2D data, which are processed efficiently using a convolutional neural network. The proposed pre-processing algorithm consists of four steps: peak detection, envelope extraction, and a down-sampling procedure. The last main step introduces the windowing with spectrum dilatation that reorders 1D data into 2D data that can be considered as an image. The spectrum dilation techniques ensure the classifier’s robustness by suppressing measurement bias, reducing the complexity of the THz dataset with negligible loss of accuracy, and speeding up the network classification. The experimental results showed that the proposed approach achieved high accuracy using a CNN classifier, and outperforms 1D classification of THz data using support vector machine, naive Bayes, and other popular classification algorithms.

The forecast of electricity demand has been a recurrent research topic for decades, due to its economical and strategic relevance. Several Machine Learning (ML) techniques have evolved in parallel with the complexity of the electric grid. This paper reviews a wide selection of approaches that have used Artificial Neural Networks (ANN) to forecast electricity demand, aiming to help newcomers and experienced researchers to appraise the common practices and to detect areas where there is room for improvement in the face of the current widespread deployment of smart meters and sensors, which yields an unprecedented amount of data to work with. The review looks at the specific problems tackled by each one of the selected papers, the results attained by their algorithms, and the strategies followed to validate and compare the results. This way, it is possible to highlight some peculiarities and algorithm configurations that seem to consistently outperform others in specific settings. Xunta de Galicia | Ref. ED431B 2020/34 Ministerio de Educación y Ciencia | Ref. TIN2017-87604-R