Search

Alertness mechanism of unmanned monitoring vehicles to environment is important. Especially, the vigilance modeling of underground security robots has a particularly significance because the underground is a dangerous environment. However, there is no a mature methodology for the alertness computation. In this work, four parts of the alertness estimation are focused. First, an autonomous robot alertness mechanism framework is proposed by using the deep reinforcement learning model of the human alertness mechanism. Second, a fast K-T filtering algorithm is developed to eliminate the multiple noises of the electroencephalograph (EEG) signals by the blind source separation and the adjustable Q factor wavelet transform. Third, the description problem of the directed interaction stability of the cortical EEG signals is solved by the ensemble empirical mode decomposition and the directional transfer function. Fourth, the human alertness estimation is explored by using the support vector regression of the dynamically spatial-temporal brain network connection parameters. Experiments show that, the mean square error and the determination coefficient of the explored alertness estimation are respectively 0.115 and 0.8337. Compared with the scalp EEG alertness estimation, it has a better performance because the mean square error is decreased by 0.0684, and the determination coefficient is increased by 0.023.

Although cloud computing is one of the promising technologies for online business services, how to schedule real-time cloud jobs with high quality of service (QoS) is still challenging current techniques. With the advancing of machine learning, deep reinforcement learning (DRL) has demonstrated its outstanding capability in dispatching time-sensitive tasks. However, the reinforced rewards in DRL are typically unavailable until the completion of the scheduling for all the jobs. Considering the fact that the trajectory of jobs in cloud is always long, current DRL-based solutions will meet challenges in finding the trajectories with high rewards, and thus would have issues such that the finally trained scheduling policy is suboptimal. To improve the problem, in this article, we propose a more advanced approach called a deep adversarial imitation reinforcement learning (AIRL) framework for scheduling time-sensitive cloud jobs. Specifically, we focus on scheduling user requests in a way to maximize job successful rate along with a significant reduction on job response time. We present the detailed design of our method, and our experimental results demonstrate that AIRL can generally outperform the existing cloud job scheduling approaches, including the DRL-based method, in the presence of different real-time workload and computing resource configurations.

Data poisoning is identified as a security threat for machine learning models. This paper explores the poisoning attack against the convolutional neural network under black-box conditions. The proposed attack is “black-box,” which means the attacker has no knowledge about the targeted model’s structure and parameters when attacking the model, and it uses “poisonous-labels” images, fake images with crafted wrong labels, as poisons. We present a method for generating “poisonous-label” images that use Enhanced Conditional DCGAN (EC-DCGAN) to synthesizes fake images and uses asymmetric poisoning vectors to mislabel them. We evaluate our method by generating “poisonous-label” images from MNIST and FashionMNIST datasets and using them to manipulate image classifiers. Our experiments demonstrate that, similarly to white box data poisoning attacks, the poisonous label attack can also dramatically increase the classification error.

With the vigorous development of Internet of Things technology, the current distribution network is developing towards the information-based and intelligent distribution Internet of Things (D-IoT). D-IoT adopts the mode of the cloud computing center and the edge cloud network working together. The edge cloud network has a large number of intelligent terminals, which can well adapt to the current sharply expanding power data scale. In order to further improve the ability of the edge network in D-IoT to process data in real time, and to maximize the quality of user experience (QoE) while minimizing energy consumption when performing computing offload, this paper proposes a dynamic non-cooperative game based edge Computing task offloading strategy, considering the dynamic nature of task generation, designed a distributed iterative optimization algorithm, which decomposes computing offloading into a series of sub-problems to solve. The results of simulation experiments prove that the calculation offloading mechanism proposed in this paper can greatly improve D -Compute efficiency of IoT system.

Background: Phasor Measurement Unit (PMU) Data Manipulation Attacks (PDMA) can change the state estimates of power systems and cause significant damage to the smart grid. So it is vital to research a method to detect it. Objective: In this paper, we propose a detection mechanism and model for PDMA. Methods: Firstly, the distribution's characteristics of Phasor Data Concentrator (PDC) and PMU are analyzed, and we use these characteristics to detect a PDMA detection mechanism that could help us reduce the number of detection samples. Secondly, we use the Sliced Recurrent Neural Network (SRNN) to extract the time series data's temporal characteristics of PMU data. Thirdly, based on the temporal characteristics, the Convolutional Neural Networks (CNN) and Attention mechanisms are used to extract the spatial features of these data. Finally, we sent the spatial features to the Fully Layer and used the softmax function to classify. Results: The proposed SRCAM in this paper has two advantages. One is that it implements the parallel computation on data by using the segmentation concept of SRNN, which reduces the computation time. The other is that using the Attention mechanism on CNN can make the spatial features more prominent. At the end of the paper, we do many comparative experiments between SRCAM and other models, such as some algorithms of Machine learning and soft computing. We take IEEE node data as experimental data and TensorFlow as an experimental platform. Experimental results show that the SRCAM model has an excellent performance of the detection of PDMA with high precision and accuracy. Conclusion: The superiority of SRCAM is theoretically and experimentally proved in this paper. As we expected, SRCAM showed great results in the application of PDMA detection.

Background As a nutritional index, preoperative serum prealbumin highly correlates with surgical complications. However, the correlation between preoperative prealbumin and postoperative complications remains unclear in liver transplantation (LT). Methods A total of 191 patients who underwent LT between 2015 and 2019 were included in the retrospective analysis. According to a cut-off value calculated from a receiver operating characteristic (ROC) curve, the patients were divided into normal and low preoperative prealbumin groups. Univariable and multivariable logistic regression analyses were used to identify independent risk factors for postoperative complications. In addition, patients were divided into subgroups by Model for End-stage Liver Disease (MELD) score, and the association between preoperative prealbumin and postoperative complications was also assessed in each group. Results A total of 111 (58.1%) patients were included in the low prealbumin group based on a cut-off value of 120 mg/L. The area under the ROC curve (AUC) was 0.754 (95% confidence interval [CI] 0.678–0.832). Low prealbumin (95% CI 1.51–12.8, P = 0.007) was identified as a predictor for postoperative complications based on multivariable regression. In the low and normal prealbumin groups, the prevalence rates of postoperative complications were 27.5% and 8.0% (P = 0.003) in the MELD score ≤ 15 subgroup and 53.3% and 20.0% (P = 0.197) in the MELD score > 15 subgroup, respectively. Conclusions Preoperative prealbumin was associated with postoperative complications in LT, and preoperative nutritional support benefitted postoperative recovery, especially for patients with low MELD scores.

This paper analyzes the influence of the global positionong system (GPS) spoofing attack (GSA) on phasor measurement units (PMU) measurements. We propose a detection method based on improved Capsule Neural Network (CapsNet) to handle this attack. In the improved CapsNet, the gated recurrent unit (GRU) is added to the front of the full connection layer of the CapsNet. The improved CapsNet trains and updates the network parameters according to the historical measurements of the smart grid. The detection method uses different structures to extract the temporal and spatial features of the measurements simultaneously, which can accurately distinguish the attacked data from the normal data, to improve the detection accuracy. Finally, simulation experiments are carried out on IEEE 14-, IEEE 118-bus systems. The experimental results show that compared with other detection methods, our method is proved to be more efficient.

In the field of image enhancement, using deep learning methods to enhance low-light images is currently mainstream. However, the methods often have complex network structures of a large number of parameters, and their training often uses paired data-sets, which are difficult to obtain in actual practice. To solve these problems, this paper proposes a simple generative adversarial network and Retinex model, dubbed RetinexGAN, that is completely trained using unpaired data-sets. It contains a decomposition network and two discriminator networks. To reduce the parameters of the network, only two convolution layers are used in the decomposition network. We show more challenging testing data where some parts of the image are underexposed and others are normal light. Both quantitative and visual results show that RetinexGAN is largely superior to state-of-the-art methods.

Background The prognostic value of myocardial trabecular complexity in patients with hypertrophic cardiomyopathy (HCM) is unknown. Purpose To explore the prognostic value of myocardial trabecular complexity using fractal analysis in participants with HCM. Materials and Methods The authors prospectively enrolled participants with HCM who underwent 3.0-T cardiovascular MRI from August 2011 to October 2017. The authors also enrolled 100 age- and sex-matched healthy participants to form a comparison group. Trabeculae were quantified with fractal analysis of cine slices to estimate the fractal dimension (FD). Participants with HCM were divided into normal and high FD groups according to the upper limit of normal reference value from the healthy group. The primary end point was defined as all-cause mortality and aborted sudden cardiac death. The secondary end point was the composite of the primary end point and readmission to the hospital owing to heart failure. Internal validation was performed using the bootstrapping method. Results A total of 378 participants with HCM (median age, 50 years; age range, 40-61 years; 207 men) and 100 healthy participants (median age, 46 years; age range, 36-59 years; 55 women) were included in this study. During the median follow-up of 33 months ± 18 (standard deviation), the increased maximal apical FD (≥1.325) had a higher risk of the primary and secondary end points than those with a normal FD (

Background: Advanced Metering Infrastructure (AMI) for the smart grid is growing rapidly which results in the exponential growth of data collected and transmitted in the device. By clustering this data, it can give the electricity company a better understanding of the personalized and differentiated needs of the user. Objective: The existing clustering algorithms for processing data generally have some problems, such as insufficient data utilization, high computational complexity and low accuracy of behavior recognition. Methods: In order to improve the clustering accuracy, this paper proposes a new clustering method based on the electrical behavior of the user. Starting with the analysis of user load characteristics, the user electricity data samples were constructed. The daily load characteristic curve was extracted through improved extreme learning machine clustering algorithm and effective index criteria. Moreover, clustering analysis was carried out for different users from industrial areas, commercial areas and residential areas. The improved extreme learning machine algorithm, also called Unsupervised Extreme Learning Machine (US-ELM), is an extension and improvement of the original Extreme Learning Machine (ELM), which realizes the unsupervised clustering task on the basis of the original ELM. Results: Four different data sets have been experimented and compared with other commonly used clustering algorithms by MATLAB programming. The experimental results show that the US-ELM algorithm has higher accuracy in processing power data. Conclusion: The unsupervised ELM algorithm can greatly reduce the time consumption and improve the effectiveness of clustering.

In recent years, with the continuous breakthrough in deep learning, convolutional neural networks (CNNs) have shown great potential for semantic segmentation. CNNs achieve better results by deepening or widening the network, but they increase the utilization rate of computing resources and even have the phenomenon of the vanishing gradient. A new convolutional neural network architecture with alternately updated clique (CliqueNet) can get a deeper network and improves the utilization of network features. In order to maximize the transmission of semantic information, this paper introduces the clique block of CliqueNet and proposes a new fully convolutional network based on the encoder-decoder structure, which calls the CyclicNet, an alternately updated network for semantic segmentation. Besides, the long skip connections and short skip connections are added in the network to avoid the vanishing gradient. The experiment was conducted on the CamVid and Cityscapes. Comparing it with the current advanced architectures shows that CyclicNet can maximize information flow and achieve the most superior results.

Aims To evaluate the prognostic value of left ventricular (LV) remodelling index (RI) in idiopathic dilated cardiomyopathy (DCM) patients. Methods and results We prospectively enrolled 412 idiopathic DCM patients and 130 age- and sex-matched healthy volunteers who underwent cardiovascular magnetic resonance imaging between September 2013 and March 2018. RI was defined as the cubic root of the LV end-diastolic volume divided by the mean LV wall thickness on basal short-axis slice. The primary endpoint included all-cause mortality and heart transplantation. The secondary endpoint included the primary endpoint and heart failure (HF) readmission. During the median follow-up of 28.1 months (interquartile range: 19.3–43.0 months), 62 (15.0%) and 143 (34.7%) patients reached the primary and secondary endpoints, respectively. Stepwise multivariate Cox regression showed that RI [hazard ratio (HR) 1.20, 95% confidence interval (CI) 1.11–1.30, P Conclusions RI is an independent predictor of all-cause mortality, heart transplantation, and HF readmission in DCM patients and provides incremental prognostic value to LVEF and LGE presence.

Big data and cloud computing could bring security problems. In order to ensure data security and user privacy, people would choose to store data in the cloud with ciphertext. How to search data eff...

This paper introduces the concept of hierarchical and zonal scheduling and proposes an iterative two-layer model to optimize the charging and discharging trading of electric vehicles (EVs), so as to minimize the overall load variance of the distribution network under the constraints of power flow and vehicle travel demand. In order to solve the mixed-integer programming (MIP) problem that exists in this model, an improved heuristic algorithm, the adaptive inertia weight krill herd (KH) algorithm is proposed. In addition, we design a decentralized trading architecture and related electricity trading process based on the consortium blockchain to ensure the security and privacy of two-way electricity trading between EVs and the smart grid. The IEEE nodes based simulation experiment shows that our scheme can effectively smooth power load fluctuations, and the improved KH algorithm can effectively improve the efficiency of model solving. Security analysis qualitatively proves that our scheme can ensure the security and privacy-preserving of electricity trading. Finally, our scheme is implemented in the Hyperledger Fabric to evaluate the feasibility and effectiveness.

Purpose: The present study was conducted to evaluate the pain-reducing efficacy and safety of the morphine and dexamethasone incorporated cocktail regimen in patients undergoing total hip arthroplasty (THA).Methods: The current study enrolled 75 patients and randomized them into three groups. The patients in group I were injected with morphine, dexamethasone, bupivacaine, flurbiprofen axetil, and normal saline. Patients assigned to group II were injected with dexamethasone, bupivacaine, flurbiprofen axetil, and normal saline. Patients in group III were given bupivacaine, flurbiprofen axetil, and normal saline. The visual analog scale (VAS) score and active and passive range of movement (ROM) were recorded and evaluated to compare the pain-controlling efficiency. The side effects of headache, dizziness, nausea, vomiting, wound leakage, and wound infection were monitored.Results: The patients in group I had the lowest VAS, and highest active ROM and passive ROM. The VAS reduced from day 1 to day 4, and the ROM increased conversely. The patients in group II presented a smaller reduction in VAS and a smaller increase in ROM, compared to the patients in Group III. The side effects of headache, dizziness, nausea, and vomiting were not recorded. Wound leakage was observed from 4 patients in group I (2 patients) and group II (2 patients). The was no statistical intergroup difference (p = 0.85). No wound infection symptoms were detected.Conclusions: Morphine and dexamethasone incorporated cocktail regimen efficiently reduced postoperative pain in patients undergoing primary total knee arthroplasty.Trial registration: Drug clinical trial registration system, CTR20181569. Registered 20 October 2018, http://www.chinadrugtrials.org.cn/clinicaltrials

The creatine transporter (CrT) maintains brain creatine (Cr) levels, but the effects of its deficiency on energetics adaptation under stress remain unclear. There are also no effective treatments for CrT deficiency, the second most common cause of X-linked intellectual disabilities. Herein, we examined the consequences of CrT deficiency in brain energetics and stress-adaptation responses plus the effects of intranasal Cr supplementation. We found that CrT-deficient (CrT–/y) mice harbored dendritic spine and synaptic dysgenesis. Nurtured newborn CrT–/y mice maintained baseline brain ATP levels, with a trend toward signaling imbalance between the p-AMPK/autophagy and mTOR pathways. Starvation elevated the signaling imbalance and reduced brain ATP levels in P3 CrT–/y mice. Similarly, CrT–/y neurons and P10 CrT–/y mice showed an imbalance between autophagy and mTOR signaling pathways and greater susceptibility to cerebral hypoxia-ischemia and ischemic insults. Notably, intranasal administration of Cr after cerebral ischemia increased the brain Cr/N-acetylaspartate ratio, partially averted the signaling imbalance, and reduced infarct size more potently than intraperitoneal Cr injection. These findings suggest important functions for CrT and Cr in preserving the homeostasis of brain energetics in stress conditions. Moreover, intranasal Cr supplementation may be an effective treatment for congenital CrT deficiency and acute brain injury.

The application of the network technology in the power grid makes the Load Frequency Control (LFC) system more vulnerable to various kinds of network attacks. The Denial of Service (DOS) attack can block the data collected by the Phasor measurement unit from being transmitted to the LFC center, thereby affecting the decision of the control center and generation of control signals, and can not adjust the frequency of the power grid timely. Aiming at the DOS attack on LFC, a defense method based on data prediction is proposed. Through the combination of the deep learning algorithm and the Extreme Learning Machine (ELM) algorithm, the Deep auto-encoder Extreme Learning Machine (DAELM) algorithm combines the advantages of the fast speed of the extreme learning machine and the advantages of high accuracy of the deep learning. We can predict and supplement the lost data based on the DAELM algorithm, and ensure the normal operation of the LFC system, thus can prevent DOS attacks. The experiments verified the effectiveness of the proposed method.

Face recognition based on deep learning has become one of the mainstream identity authentication technologies. In recent years, many well-developed deep convolutional neural networks have emerged. But most of these network structures are very complicated, the training processes are very difficult and face recognition based on these network structures are too large to apply to mobile terminals. To solve this problem, we propose a lightweight face recognition algorithm: LightFace, which is based on depthwise separable convolution. In the process of training LightFace, the triplet loss algorithm is used to optimize model. However, LightFace requires a large amount of training data and the parameters of this model will record the users' data, which can be recovered by attackers. To address this problem, we propose an applicable approach to providing strong privacy guarantees for LightFace. In our approach, the generated data and additional noise are used to slightly disturb the original data distribution, so that the attackers cannot correctly predict the training data, which can improve the security of the model. Besides, in the process of training LightFace, ensemble learning increases the randomness of the original data distribution and enhances the robustness of the model. Within the differential privacy framework, we analyzed the privacy loss of the algorithm theoretically and conduct experiments on different face recognition datasets to demonstrate the effectiveness of our privacy preservation method. The experiment results show that LightFace with privacy preservation still has good recognition accuracy while protecting data privacy.

Currently, the booming popularity and growth of mobile devices in urban cities leads to the surge of various computation-intensive mobile applications, such as virtual reality and online video, which are strict with the computing capability and battery life of the mobile devices. To address this issue, smart edge computing in Wireless Metropolitan Area Networks (WMAN) is proposed to enable mobile users to offload computation-intensive tasks to the edge computing nodes which deploys computing resources nearby the mobile devices. However, the normal operation of edge computing nodes consumes plenty of energy. Thus, it is still a challenge to be aware of energy consumption while the computing tasks are migrated to the Edge Computing Nodes (ECNs). In view of this challenge, an Energy-Aware Computation Offloading method, named EACO, is designed to reduce the energy consumption. Technically, we analyze all access point (AP) routings between the original AP to destination AP and select the shortest path to offload the computing tasks. Furthermore we adopt Non-dominated Sorting Genetic Algorithm II (NSGA-II) to realize multi-objective optimization to shorten the offloading time of the computing tasks and reduce the energy consumption of the ECNs. Besides, we exploit Multiple Criteria Decision Marking (MCDM) and Simple Additive Weighting (SAW) to select the optimal offloading solution. Finally, the simulation experimental results show that our proposed EACO outperforms other methods.

Remarkable theoretical and experimental achievements of quantum technology have convincingly shown that a flourishing quantum age is coming. The threat of quantum attacks posed to the classical cryptographic protocols, whose security lies on the computational complexity of difficult problems, is ever more severe. As a particular voting and typical daily activity, the anonymous veto has already been further studied in classical cryptography. In this paper, a novel quantum anonymous veto protocol with information-theoretic security is proposed. Benefited from the entanglement of GHZ states and quantum nonlocality, our protocol satisfies some desirable properties of reliability, privacy, verifiability, tracelessness, and fairness. Compared to the properties of classical protocols, tracelessness is a unique property in quantum anonymous veto protocol. Using the online analog quantum computer of IBM Corporation placed on the cloud, we make an experimental test of anonymous veto for four voters. Any harmful attempt of cheating from the semi-honest certificate authority center, malicious voters, and external eavesdroppers will be detected by honest participants based on the fantastic quantum world. This work to voting theory helps elucidate how quantum mechanics can be harnessed for secure strategic advantage.

The edge perception layer of electric Internet of Things (eIoT) involves a large amount of user’s data, including identity, location and other sensitive data, facing the risk of privacy disclosure. To solve this problem, we first established an eIoT model based on edge IoT agent, and propose a dynamic privacy preserving scheme using Multi-identity-based Fully Homomorphic Encryption (MIBFHE). Different from the exiting schemes, the system is no longer limited to the same secret key. Each user generating their own secret key on the basis of identity, and key management, revocation are much easier to achieve. In addition, our scheme is dynamically faulttolerant, and it can still be normal to calculate even if the device is added or failed. The experimental results verify the effectiveness and advantage in the cost of computational and communication compared to the existing schemes.

Smart grid is a highly integrated power system that combines modern advanced information and communication, sensor measurement and automatic control technology, including a large number of smart devices, with reliability, flexibility, safety and self-healing advantages a great service. However, devices such as smart meter in the network face the issue of disclosing user privacy when collecting, processing, and transmitting large amounts of data. To solve this problem, this paper puts forward a scheme based on improved group signature and homomorphic encryption, using improved group signature to determine the public and private key of smart meter and the signature of electricity data, encrypting and aggregating the electricity data with homomorphic encryption. Finally, in the transmission of data, the electricity data is always in the form of ciphertext, and the group signature scheme with forwarding security can avoid the harm caused by the disclosure of the private key, thus playing a role in privacy-preserving.

Brain computer interface (BCI) technology refers to the technology that helps establish direct communication and control channels between the human brain and the computer. This technology can meet the control requirements of external equipment without passing through the patient's damaged nerves and declining muscle system. Electroencephalogram (EEG) contains a variety of information, including motor imagination information, steady-state visually evoked EEG information, emotional information, and fatigue information, etc. This paper designs a brain-computer interface system based on motor imagination and uses it to control the body movements of the NAO robot. The system is mainly divided into three modules: signal acquisition, signal processing and robot control. In terms of signal acquisition, electrode caps, conductive glue, EEG amplifiers and Curry7 software are used. In this paper, we design a frequency band feature selection algorithm based on an improved sub-band filter bank (FB)-Common Space Pattern (CSP). These features are classified using Bagging ensemble classification algorithm and converted to use Commands for controlling the NAO robot. In terms of communication control, the instruction codes are preset, and the classification results are corresponding to instructions for controlling different actions of the robot. It is transmitted to the robot by using wireless communication technology. The Bagging ensemble classification method proposed in this paper classifies EEG features. The accuracy of the four types of motor imaging EEG classification is as high as 78.29%, and the accuracy of NAO robot action execution experiments is up to 86.33%, which proves the correctness of the algorithm studied in this paper. The reliability and feasibility of the NAO robot brain control system designed in this paper.

Based on the convolutional neural network (CNN) and long short-term memory neural network (LSTM), combined with data enhancement technology and spatial-temporal attention mechanism, a scene recognition model is established. In order to balance the difference in the amount of data between different samples, data enhancement technology based on video data samples is introduced. Aiming at improving the performance of the model, a spatial-temporal attention mechanism is used to improve the accuracy of scene recognition. The model sample scenes used in this paper include three types: fully mechanized coal face, coal mine roadway and mining machinery. The experimental verification shows that: compared with the traditional convolutional neural network, the accuracy of the CNN-LSTM model is improved by 2.136%, and the CNN-LSTM model with spatial-temporal attention mechanism is improved by 2.9210/0. The accuracy of the deep CNN-LSTM model with spatial-temporal attention mechanism is about 93.0630/0.

With the growth of volume of high quality information in the Deep Web, as the key to utilize this information, Deep Web data source classification becomes one topic with great research value. In this paper, we propose a Deep Web data source classification method based on text feature extension and extraction. Firstly, because the data source contains less text, some data sources even contain less than 10 words. In order to classify the data source based on the text content, the original text must be extended. In text feature extension stage, we use the N-gram model to select extension words. Secondly, we proposed a feature extraction and classification method based on Attention-based Bi-LSTM. By combining LSTM and Attention mechanism, we can obtain contextual semantic representation and focus on words that are closer to the theme of the text, so that more accurate text vector representation can be obtained. In order to evaluate the performance of our classification model, some experiments are executed on the UIUC TEL-8 dataset. The experimental result shows that Deep Web data source classification method based on text feature extension and extraction has certain promotion in performance than some existing methods.

The rise of the smart grid, which applies the Internet and smart metering technologies to the power system, results in massive, heterogeneous power data and a large number of attacks on the smart grid and causes great threats to the confidentiality and integrity of the power data. At the same time, with the development of quantum information, it is possible to use quantum cryptography to protect data transmission. Aiming at the problem of data security transmission in smart grid, this paper proposes a lightweight transmission scheme for power data security. In the scheme, we combine quantum cryptography with ”one-time pad” mechanism, use a quantum random number generator which compensates the defects of the traditional random number generator and quantum key distribution protocol which ensure the security of key distribution, and design a new lightweight stream cipher encryption algorithm. Through theoretical analysis and experiments, it is proved that the proposed transmission scheme can ensure the security of power data transmission.

Bidirectional communication infrastructure of smart systems, such as smart grids, are vulnerable to network attacks like distributed denial of services (DDoS) and can be a major concern in the present competitive market. In DDoS attack, multiple compromised nodes in a communication network flood connection requests, bogus data packets or incoming messages to targets like database servers, resulting in denial of services for legitimate users. Recently, machine learning based techniques have been explored by researchers to secure the network from DDoS attacks. Under different attack scenarios on a system, measurements can be observed either in an online manner or batch mode and can be used to build predictive learning systems. In this work, we propose an efficient DDoS attack detection technique based on multilevel auto-encoder based feature learning. We learn multiple levels of shallow and deep auto-encoders in an unsupervised manner which are then used to encode the training and test data for feature generation. A final unified detection model is then learned by combining the multilevel features using and efficient multiple kernel learning (MKL) algorithm. We perform experiments on two benchmark DDoS attack databases and their subsets and compare the results with six recent methods. Results show that the proposed method outperforms the compared methods in terms of prediction accuracy.

In order to realize peer-to-peer (P2P) transactions between electric vehicles (EVs) in vehicle-to-grid (V2G) networks, we propose an EV power trading model based on blockchain and smart contract. Firstly, based on the blockchain and smart contract technology, a decentralized power trading model is proposed to realize the information equivalence and transparent openness of power trading. Then, considering the randomness and uncertainty of EV charging and discharging, the EV trading parties use the reverse auction mechanism based on dynamic pricing strategy to complete the transaction matching, which can not only improve the profit of the less competitive power seller, but also it can reduce the cost of the electricity purchaser. Finally, in order to verify the feasibility of our proposed scheme, V2G's EV power trading smart contract was designed, and the smart contract was released to Ethereum and simulated experiments were carried out. The effectiveness of the proposed scheme is verified by simulation experiments and comparison with traditional power trading schemes.

Background: Coronavirus disease 2019 (COVID-19) is a highly infectious disease. A small proportion of discharged patients may become positive again for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA, even if they meet the discharge criteria. Herein, we report a rare COVID-19 patient with recurrent recurrence of positive SARS-CoV-2 RNA.Case presentation: A 68-year-old man was admitted due to fever, muscle pain, and fatigue. He was initially diagnosed with COVID-19 according to two consecutive positive results for SARS-CoV-2 RNA plus clinical symptoms and chest CT findings, and was discharged from hospital when meeting the discharge criteria, including two consecutive negative results. He was tested positive for SARS-CoV-2 RNA twice during the quarantine and was hospitalized again. He was asymptomatic then, but IgG and IgM were both positive. He was discharged in the context of four consecutive negative test results for SARS-CoV-2 RNA after antiviral treatment. However, he was tested positive once again on the 3rd and 4th day after the second discharge, although still asymptomatic. IgG and IgM were still positive. After antiviral treatment, the results of SARS-CoV-2 RNA were negative in three consecutive retests, and he was finally discharged and quarantined for further surveillance. Conclusion: This case suggests that a small proportion of convalescent patients may become positive again for SARS-CoV-2 RNA and be a virus carrier.

In recent years, various types of power theft incidents have occurred frequently, and the training of the power-stealing detection model is susceptible to the influence of the imbalanced data set and the data noise, which leads to errors in power-stealing detection. Therefore, a power-stealing detection model is proposed, which is based on Improved Conditional Generation Adversarial Network (CWGAN), Stacked Convolution Noise Reduction Autoencoder (SCDAE) and Lightweight Gradient Boosting Decision Machine (LightGBM). The model performs Generation- Adversarial operations on the original unbalanced power consumption data to achieve the balance of electricity data, and avoids the interference of the imbalanced data set on classifier training. In addition, the convolution method is used to stack the noise reduction auto-encoder to achieve dimension reduction of power consumption data, extract data features and reduce the impact of random noise. Finally, LightGBM is used for power theft detection. The experiments show that CWGAN can effectively balance the distribution of power consumption data. Comparing the detection indicators of the power-stealing model with various advanced power-stealing models on the same data set, it is finally proved that the proposed model is superior to other models in the detection of power stealing.

Background: As the sixth most common cancer of worldwide, head and neck cancers (HNC) are springing from oral cavity, pharynx and larynx and there is no strong biomarker for prognosis. Rates of 5 years survival with HNC remain relatively low in decades with improvement of treatments. Evidence that single nucleotide polymorphisms (SNPs) play a part in cancer prognosis is growing. Methods: We conducted an exome-wide association study among 261 patients with head and neck squamous cell carcinoma (HNSCC) and then validated in The Cancer Genome Atlas (TCGA) database for survival by using the Cox proportional hazards regression models and Kaplan–Meier analyses. Results: After combining the result of the two stages, 4 SNPs were significantly associated with HNSCC survival (rs16879870 at 6q14.3: adjusted HR = 2.02, 95%CI = 1.50–2.73, P = 3.88 × 10−6; rs2641256 at 17p13.2: adjusted HR = 0.67, 95%CI = 0.56–0.80, P = 7.51 × 10−6; rs2761591 at 11p13: adjusted HR = 2.07, 95%CI = 1.50–2.87, P = 1.16 × 10−5; and rs854936 at 22q11.21: adjusted HR = 1.92, 95%CI = 1.43–2.57, P = 1.27 × 10−5). Besides, we constructed a receiver operating characteristic (ROC) model to estimate predictive effect of the novel SNPs combined with clinical stage in HNSCC prognosis (AUC = 0.715). We also found the genotype of rs16879870 and rs854936 was significantly associated with the expression of gene GJB7 (P = 0.013) and RTN4R (P = 0.047) in cancer tissues of TCGA, respectively. Conclusion: Our findings suggested that the SNPs (rs16879870, rs2641256, rs2761591, rs854936) might play a crucial role in prognosis of HNSCC.

We read with interest the article by Ferro et al.1 that focused on acute cerebral microinfarcts (ACMIs) in vascular cognitive impairment.

Currently, cloudlet based wireless metropolitan area network (WMAN) is emerging as an effective paradigm to extend the performance of mobile devices, which enables the execution of computational intensive mobile applications. But the normal operation of cloudlets consumes a large amount of energy, which brings about carbon dioxide emissions, aggravation of the greenhouse effect, etc. Meanwhile the data transmission of various mobile users among cloudlets would cause leakage of personal privacy. In view of this challenge, we propose an energy-efficient cloudlet management method, named ECM, for privacy preservation in WMAN. Technically, an optimization model is designed to formalize our problem. Then, based on live virtual machine (VM) migration technique, a corresponding privacy-aware VM scheduling method for energy saving is designed to determine which VMs should be migrated and where they should be migrated. Finally, experimental data demonstrate that the proposed method is both efficient and effective.

DBN- based classification model is proposed in this paper for web spam detection.SMOTE and DAE algorithms is applied in DBN to improve the classification performanceThe content and link feature is combined.The results obtained in this paper is better than the existing systems. With the development of the Internet, the number of web spam increases gradually, which has seriously affected the user experience of search engines. To improve the classification performance of web spam, the deep belief networks (DBN) is used for the first time, and it is effectively combined with the Synthetic Minority Over-Sampling Technique (SMOTE) and De-Noising Auto-Encoder (DAE) algorithm after the multi-aspect research and consideration. After multiple sets of experiments on WEBSPAM-UK2007 dataset, the results show that the classification method proposed in this paper improves the classification performance to a certain extent, which provides a good direction for the future classification of web spam.

Nowadays, identity protection has turned into a fundamental demand for online activities. Currently, the present quantum anonymous communication protocols mostly rely on multi-entanglement. In this paper, we propose an anonymous communication protocol for anonymous sender by using single-particle states. The protocol can be extended to a communication protocol where the sender and receiver are fully anonymous with the message kept secret. In terms of security, our protocol is designed to comply with the technique of collective detection. Compared to the step-by-step detection, collective detection, in which the participants perform detection only once, reduces the complexity of the protocol to some extent. Moreover, we analytically demonstrate the security of the protocol in the face of active attacks. Any active attack employed by an external or internal attacker cannot reveal any useful information about the sender’s identity. Meanwhile, any malicious behavior will be detected by honest participants.

The patient with a severely spinal injury needs an intention-controlled nursing bed (NB) robot to help the body actions. To recognize human intentions, the brain–computer interface (BCI) of the steady-state visual evoked potentials (SSVEPs) is usually used. For the BCI of the SSVEPs, the visual fatigue should be alleviated, and the intention recognition accuracy should be improved. Therefor this paper proposed a coloring and timing BCI (CT-BCI) to alleviate the visual fatigue, and find the optimal Wavelet vanishing moment parameter to improve the intention recognition accuracy. In addition, the intention coding and decoding are designed and implemented. Furthermore, the wireless communications are used among the CT-BCI and the robots in the internet of things. Experiments show that the proposed CT-BCI decreases the average evocation time for each intention by 0.8 s, and prolongs the eye non-fatigue duration by 5.45 s, and improves the intention recognition accuracy by 6.7% compared with the traditional BCI.

In view of the low accuracy of the traditional emotion recognition methods based on facial expressions, an emotion recognition method based on maximum weight multi-modal information fusion of electroencephalographs (EEGs) and facial expression information is proposed in this paper. First, the induced emotional EEG data is converted into the corresponding EEG topographic map data and sent to the convolutional network for training and outputting decision information. Second, the illumination compensation method is utilized to filter the noise of the face image data. Then, the face image data is trained in the multi-scale feature extraction network, and the decision information is output. Finally, aiming at the decision-level information fusion, a weighted fusion method is proposed in this paper for emotion recognition. Experimental tests show that the recognition accuracy of the multi-scale feature extraction network on the CK+ data set and Fer2013 data reached 94.4% and 72%, respectively. Simultaneously, the multi-modal information fusion method achieves 92.6% accuracy in emotion recognition.

Power grid is a complex system which closely links the power generation and power consumer through transmission and distribution networks. With the development of smart grid, smart grid is more open to external communication systems, it also has exposed some problems in the network attacks. A new false data injection attack (called the unobservable attack) that can bypass the traditional BDD and inject random errors into state estimation. We propose an improved extreme learning machine (ELM) for attack detection. The artificial bee colony (ABC) incorporates the thought of differential evolution algorithm (DE) to optimize ELM for improving detection precision. In this paper, Autoencoder is used to reduce the dimensionality of the measurement data, which makes the low-dimensional data information basically and fully represent high-dimensional data. We verify the performance of the proposed method on IEEE bus systems, and prove that the proposed method can effectively detect such unobservable attack.

Reactive power optimization is closely related to voltage quality and network loss, and it has great significance for the safety, reliability, and economical operation of the power system. Differential evolution (DE) algorithm has been currently applied to reactive power optimization. In order to mitigate the shortcomings of poor local search ability and premature convergence in DE, this paper presents a novel hybrid algorithm–chaotic artificial bee colony differential evolution (CABC-DE) algorithm, which improves the DE algorithm based on artificial bee colony algorithm and ideas of chaotic search. It introduces the observation bees' acceleration operation and the detective bees' chaotic search operation into CABC-DE. The validity of the proposed method is examined using IEEE-14 and IEEE-30 bus system. The experimental results show that CABC-DE algorithm is more effective than regular DE algorithm for reactive power optimization. The algorithm can save the search time greatly and get a better sol...

To investigate the prognostic significance of T1 mapping using T1 long and short in hypertrophic cardiomyopathy (HCM) patients. A total of 263 consecutive patients with HCM referred for cardiovascular magnetic resonance (CMR) imaging were enrolled in this study. The imaging protocol consisted of cine, late gadolinium enhancement (LGE), and T1 mapping with T1 long and short. All patients were followed up prospectively. Outcome events were divided into the primary and secondary endpoint events. Primary endpoint events included cardiac death, heart transplant, aborted sudden death, and cardiopulmonary resuscitation after syncope. The secondary endpoint event was defined as unplanned rehospitalization for heart failure. The average follow-up duration was 28.3 ± 12.1 (range: 1–78) months. In all, 17 patients (7.0%) experienced a primary endpoint including 13 cardiovascular deaths, three aborted sudden deaths, and one resuscitation after syncope, and 34 patients experienced a secondary endpoint. Patients with primary endpoints showed a trend towards more extensive LGE (p < 0.001), significantly higher ECV (p < 0.001), and native T1 (p = 0.028) than those without events. In multivariate Cox regression analysis, ECV was independently associated with primary and secondary endpoints (p < 0.001 and p = 0.047, respectively). For every 3% increase, ECV portended a 1.374-fold increase risk of a primary endpoint occurring (p < 0.001). In the Kaplan–Meier survival analysis, the incidence of primary and secondary endpoint events was significantly higher in HCM with increased ECV (p < 0.001 and p = 0.009, respectively). In patients with HCM, ECV is a strong imaging marker for predicting adverse outcome. • ECV is a potent imaging index which has a strong correlation with LVEF and LVEDVI and can evaluate myocardial tissue structure and function. • ECV and LGE can provide a prognostic value in patients with hypertrophic cardiomyopathy. • ECV has stronger predictive effectiveness than LGE; even in the subgroup with LGE, ECV shows independent predictive significance for adverse events.

The false data injection attack can tamper with the measurement information collected by the Supervisory Control and Data Acquisition, threat the security of state estimation. Therefore, the analysing methods and detection methods of false data injection attacks have important theoretical and practical significance for ensuring the safe operation of smart grids. This study uses the invertible automatic encoder to reduce the dimension of the original data and uses the long-short-term memory to detect false data injection attacks. This method overcomes the shortcomings of shallow algorithm and traditional machine learning algorithm for power big data training and avoiding the problems of gradient explosion and gradient disappearing during training. Finally, the authors perform a large number of experiments in the IEEE 118-node test system and the 300-node test system and verify the effectiveness of the proposed method.

Basilar artery (BA) dolichosis is not uncommon in patients with acute isolated pontine infarction. The effect of this abnormal BA geometrical form on the outcomes of pontine infarction has not been closely examined. This study aims to elucidate whether BA dolichosis contributes to a poor 90-day outcome in acute isolated pontine infarction. A total of 101 patients were enrolled with a median age of 65 years. The BA diameter (p = 0.026), basilar artery length (BAL) (p p p p p = 0.042) as well as the proportions of BA dolichosis (p = 0.007) and BA bending (p = 0.010) in univariate analysis. Multivariate logistic regression analysis determined that BA dolichosis (adjusted OR = 4.724, 95% CI: 1.481~15.071, p = 0.009) and baseline NIHSS score (adjusted OR = 1.805, 95% CI: 1.296~2.513, p

Background Long axis strain (LAS) is a parameter derived from standard cardiovascular magnetic resonance imaging. However, the prognostic value of biventricular LAS in hypertrophic cardiomyopathy (HCM) is unknown. Methods Patients with HCM (n = 384) and healthy volunteers (n = 150) were included in the study. Left ventricular (LV)-LAS was defined as the percentage change in the length measured from the epicardial border of the LV apex to the midpoint of a line connecting the mitral annulus at end-systole and end-diastole. Right ventricular (RV)-LAS represented the percentage change of length between epicardial border of the LV apex to the midpoint of a line connecting the tricuspid annulus at end-systole and end-diastole. The primary endpoint was a combination of all-cause death and sudden cardiac death aborted by appropriate implantable cardioverter-defibrillator discharge and cardiopulmonary resuscitation after syncope. The secondary endpoint was a combination of the primary endpoint and hospitalization for congestive heart failure. Results Twenty-nine patients (7.6%) achieved the primary endpoint, and the secondary endpoint occurred in 66 (17.2%) patients. In multivariate Cox regression analysis, RV-LAS was an independent prognostic factor for the primary (hazard ratio (HR), 1.13) and secondary (HR, 1.11) endpoints. In the subgroup of patients with a normal RV ejection fraction (EF) (>45.0%, n = 345), impaired RV-LAS was associated with adverse outcomes and might add incremental prognostic value to RVEF and tricuspid annular plane systolic excursion (TAPSE) (p Conclusions RV-LAS is an independent predictor of adverse prognosis in HCM in addition to RVEF and TAPSE.

Advanced Metering Infrastructure(AMI) is a critical core component in smart grid. Currently, smart grid employs the same computer networks, which is vulnerable to suffer from cyber attacks. Furthermore, applications of traditional intrusion detection methods are restricted due to the limited computing capacity and potential deployment costs of electrical equipments. This paper proposes an ELM-based intrusion detection method for AMI. We first filter and partition the malicious data, and then different types of invasion are effectively extracted. Finally, we can use Extreme Learning Machine(ELM) for detecting different attack types of malicious data. However, traditional machine learning algorithms such as Support Vector Machine(SVM), which results in a longer training time and poor performance, and moreover SVM is not applicable to multi-class problems. In theory, ELM can approximate any target continuous function and classify any disjoint rejoins. As verified by the simulation results, ELM tends to have better scalability and achieve much better generalization performance at much faster learning speed than traditional SVM.

Context Software similarity comparison has always been a common technique for software reuse detection, plagiarism detection, and defect detection. Objective Considering the role of API calls and arithmetic operations in software execution, a semantic-based dynamic software analysis method–mixed key instruction sequence (MKIS) is proposed. Method MKIS embeds key value sets into a vector and constructs a novel software execution sequence that contains API calls and arithmetic operations during software execution. To determine the location of key values, a key-value equivalent matching algorithm is proposed, combined with the longest common subsequence algorithm to optimize the software execution sequence. Results Experiments show that MKIS can accurately compare the similarity of binary programs without obtaining the software source code, and has better resiliency and credibility. Conclusion Moreover, in the case when the software source code is changed with some main function-independent modification and code obfuscator, software reuse can be successfully detected.

We read, with interest, the article by Hurford et al.1 that focused on the long-term prognostic value of “tissue-based” definitions of TIA and minor stroke. Diffusion-weighted imaging (DWI)–positive lesions are associated with longer duration of ischemia and are also time dependent.2 Accordingly, there was a high risk of a false negative rate because of the delayed scan time in this study. The differences in the time interval of event to MRI between the DWI-positive and DWI-negative groups might illustrate this point as well, which would inevitably increase statistical discrepancy. Moreover, DWI lesions can be transient, especially if normal blood flow is restored by pharmacologic mechanisms, such as thrombolysis or antithrombotic therapy.3 Consequently, therapeutic measures should be also be taken into account.

Background We performed a systematic review and meta-analysis to assess the discrimination performance of the 2014 European Society of Cardiology (ESC) sudden cardiac death (SCD) risk-prediction model for hypertrophic cardiomyopathy (HCM). Methods We searched the PubMed, Embase and Web of Science, CNKI, WanFang, and China Biology Medicine disc (CBMdisc) databases for English and Chinese articles validating the risk model. The model's discrimination performance with cutoff points of 4% and 6% based on extracted information was calculated. The extracted C statistic and calculated area under the curve (AUC) with 95% confidence intervals (CIs) of all studies were weighted and summarized. Heterogeneity was quantified through I2 statistics; sensitivity analysis and publication bias were assessed with Egger's test. Results We included 13 studies validating the model's usefulness. We concluded that the model has excellent specificity, although it has poor sensitivity when setting a recommended cutoff value of 6% for identifying high-risk patients with HCM. In addition, there was moderate discrimination value (global C statistic = 0.75, 95% CI, 0.67-0.83; cutoff point of ≥ 4%; AUC = 0.69, 95% CI, 0.62-0.75; cutoff point of ≥ 6%; AUC = 0.65, 95% CI, 0.59-0.72). Subgroup analysis by region showed slightly weaker predictive ability for North America. There was no significant publication bias (all P > 0.05); sensitivity analysis did not change the results significantly. Conclusions The 2014 ESC HCM SCD risk-prediction model has excellent specificity and poor sensitivity and has moderate discrimination performance. In addition, it may have lower prediction value for North America compared with other regions.