Search

In the global precipitation measurement (GPM) era, the integrated multisatellite retrievals for GPM (IMERG) stands as a pivotal precipitation algorithm. This study aims to evaluate the accuracy of IMERG in capturing precipitation frequency and volume in the Yellow River Basin, China. Four satellite-based precipitation estimates (SPEs) from three algorithmic versions of IMERG were analyzed on an hourly scale using gauge observations. An improved error component procedure was employed to identify error sources. Results showed that all four IMERG products effectively captured the spatial distribution patterns and seasonal changes of precipitation, IMERG_F demonstrated the best overall performance, followed by IMERG_L, while IMERG_E performed the worst. However, they tended to overestimate precipitation. IMERG_F_Cal performed the best for precipitation frequency detection, with the highest probability of detection (POD = 52.7%) and the lowest missed events (MIS = 47.3%). Error component analysis highlights false bias as the main source of error, followed by missed bias. In winter, missed bias was the primary error source. Notably, a significant overestimation of precipitation was observed along the Yellow River. In detail, false bias dominated IMERG_E, IMERG_L, and IMERG_F_UnCal below 800 m in spring, summer, and autumn. However, in winter, missed bias became the primary error source for these three products at elevations above 200m and for IMERG_F_Cal above 500 m. IMERG_F_Cal exhibited false bias as the primary error source in all seasons. Suggested algorithm developers focus on improving IMERG SPEs' identification capabilities for light precipitation events and rainstorms. Findings can provide a reference for improving the IMERG product algorithms and enhancing users’ understanding of the error characteristics and sources of IMERG products.

Electroencephalography (EEG)-based motor imagery (MI) is one of brain computer interface (BCI) paradigms, which aims to build a direct communication pathway between human brain and external devices by decoding the brain activities. In a traditional way, MI BCI replies on a single brain, which suffers from the limitations, such as low accuracy and weak stability. To alleviate these limitations, multi-brain BCI has emerged based on the integration of multiple individuals’ intelligence. Nevertheless, the existing decoding methods mainly use linear averaging or feature integration learning from multi-brain EEG data, and do not effectively utilize coupling relationship features, resulting in undesired decoding accuracy. To overcome these challenges, we proposed an EEG-based multi-brain MI decoding method, which utilizes coupling feature extraction and few-shot learning to capture coupling relationship features among multi-brains with only limited EEG data. We performed an experiment to collect EEG data from multiple persons who engaged in the same task simultaneously and compared the methods on the collected data. The comparison results showed that our proposed method improved the performance by 14.23% compared to the single-brain mode in the 10-shot three-class decoding task. It demonstrated the effectiveness of the proposed method and usability of the method in the context of only small amount of EEG data available.

Donor-Acceptor (D-A) polymer field-effect transistors (pFETs) are at the cutting edge of organic electronics. However, some fundamental cognition of interface states remains unquantified, particularly at different dynamic scales. In this study, the interface states of D-A polymer transistors are quantified using the dynamic pumping method. The experimental results indicate that the interface state density of the transistor is insensitive to the measurement pulse condition and stays within the range of $ {10}^{ {12}}\,\,\sim \,\, {10}^{ {13}}\,\,\rm {cm}^{ {-2}}$ . The experiments described in this paper provide a quantitative analysis of interface states. This analysis can serve as effective guidance for optimizing future devices.

Background: In the past few years, U-Net based U-shaped architecture and skip-connections have made incredible progress in the field of medical image segmentation. U2-Net achieves good performance in computer vision. However, in the medical image segmentation task, U2-Net with over nesting is easy to overfit. Purpose: A 2D network structure TransU2-Net combining transformer and a lighter weight U2-Net is proposed for automatic segmentation of brain tumor magnetic resonance image (MRI). Methods: The light-weight U2-Net architecture not only obtains multi-scale information but also reduces redundant feature extraction. Meanwhile, the transformer block embedded in the stacked convolutional layer obtains more global information; the transformer with skip-connection enhances spatial domain information representation. A new multi-scale feature map fusion strategy as a postprocessing method was proposed for better fusing high and low-dimensional spatial information. Results: Our proposed model TransU2-Net achieves better segmentation results, on the BraTS2021 dataset, our method achieves an average dice coefficient of 88.17%; Evaluation on the publicly available MSD dataset, we perform tumor evaluation, we achieve a dice coefficient of 74.69%; in addition to comparing the TransU2-Net results are compared with previously proposed 2D segmentation methods. Conclusions: We propose an automatic medical image segmentation method combining transformers and U2-Net, which has good performance and is of clinical importance. The experimental results show that the proposed method outperforms other 2D medical image segmentation methods. Clinical Translation Statement: We use the BarTS2021 dataset and the MSD dataset which are publicly available databases. All experiments in this paper are in accordance with medical ethics.

For a class of discrete-time bidirectional associative memory neural networks (DTBAMNNs) with multiple time-varying delays, the issue of state estimation is studied. By propose a mathematical induction method, we first investigate novel delay-dependent and -independent global exponential stability (GES) criteria of the error system. The obtained GES criteria are described by linear scalar inequalities. Then, a state observer is derived via the theory of generalized matrix inverses. These exponential stability conditions are very simple, which is convenient to verify based on the standard software tools (for example, YALMIP). Finally, we present two illustrative examples to present the effectiveness of the theoretical results.

In practical applications, the raw input to a Knowledge Based Question Answering (KBQA) system may vary in forms, expressions, sources, etc. As a result, the actual input to the system may contain various errors caused by various noise in raw data and processes of transmission, transformation, translation, etc. As a result, it is significant to evaluate and enhance the robustness of a KBQA model to various noisy questions. In this paper, we generate 29 datasets of various noisy questions based on the original SimpleQuestions dataset to evaluate and enhance the robustness of a KBQA model, and propose a model which is more robust to various noisy questions. Compared with traditional methods, the main contribution in this paper is that we propose a method of generating datasets of different noisy questions to evaluate the robustness of a KBQA model, and propose a KBQA model which contains incremental learning and Mask Language Model (MLM) in the question answering process, so that our model is less affected by different kinds of noise in questions and achieves higher accuracies on datasets of different noisy questions, which shows its robustness. Experimental results show that our model achieves an average accuracy of 78.1% on these datasets and outperforms the baseline BERT-based model by an average margin of 5.0% with the similar training cost. In addition, further experiments show that our model is compatible with other pre-trained models such as ALBERT and ELECTRA.

Because of the undesired fatigue-related consequences, accumulating efforts have been made to find an effective intervention to alleviate the suboptimal cognitive function caused by mental fatigue. Nonetheless, limitations of intervention and evaluation methods may hinder the revealing of underlying neural mechanisms of fatigue recovery. Through the newly-developed dynamic functional connectivity (FC) analysis framework, this study aims to investigate the effects of two types of mid-task interventions (i.e., rest-break and moderate-intensity exercise-break) on the dynamic reorganization of FC during the execution of psychomotor vigilance test (PVT). Using a sliding window approach, temporal brain networks within each frequency band (i.e., $\delta $ , $\theta $ , $\alpha $ , & $\beta $ ) were estimated before and immediately after the intervention, and towards the end of the task to investigate the immediate and delayed effects respectively during post-break task reengagement. Behaviourally, similar beneficial effects of exercise- and rest-break on performance were observed, manifested by the immediate improvements after both interventions and a long-lasting influence towards the end of tasks. Moreover, temporal brain networks assessment showed significant immediate decreases of fluctuability, which was followed by an increase of fluctuability towards the end of intervention tasks. Furthermore, the temporal nodal measure revealed the channels with significant differences across tasks were mainly resided in the fronto-parietal areas that exhibited interesting frequency-dependent distribution. The observations of immediate and delayed dynamic FC reorganizations extend previous fatigue-related intervention and static FC studies, and provide new insight into the dynamic characteristics of FC during post-break task reengagement.

A novel hybrid superconducting coupling (NHSC) is proposed in this paper, in which the secondary is composed of copper and a split superconductor. A magnetic equivalent circuit (MEC) model is presented to analyze and calculate the distribution of the magnetic field and torque characteristics. The 3D modeling considers the end effects encountered in mating. The expressions of the magnetic flux and torque are obtained by combining Kirchhoff’s law and Ampere’s loop law, respectively. A 3D finite element (FE) model is built, and the simulation is performed. Its performance is investigated using a 135K critical high-temperature superconductor to have a greater operating margin of the coupling. For a high slip speed of 1500 rpm, we observed a peak torque of 167 Nm and stability of 115 Nm. By varying the thickness of the air gap or the PMs, we observed the increasing or decreasing of the torque. Due to the cryogenics systems, we also observed the reduction of the losses by the joule effect. The maximum error between the results of the torque obtained by simulation and the results calculated using the proposed model is 6.74%, which proves that both results are in good agreement.

Due to the long train marshaling and complex line conditions, the operating modes in heavy haul rail systems frequently change when trains travel. Improper traction or braking operation made by drivers will increase the longitudinal impact force to trains and causes the train decoupling, severely affecting the safe operations of trains. It is quite desirable to replace the manual control with intelligent control in heavy haul rail systems. Traditional machine learning-based intelligent control methods suffer from insufficient data. Due to lacking effective incentives and trust, data from different rail lines or operators cannot be shared directly. In this paper, we propose an approach on blockchain-based federated learning to implement asynchronous collaborative machine learning between distributed agents that own data. This method performs distributed machine learning without a trusted central server. The blockchain smart contract is used to realize the management of the entire federated learning. Using the historical driving data collected from real heavy haul rail systems, the learning agent in the federated learning method adopts a support vector machine (SVM) based intelligent control model. To deal with the imbalanced traction and braking data, we optimize the classic SVM model via assigning different penalty factors to the majority and minority classes. The data set are mapped to a high dimension using kernel functions to make it linearly separable. We construct a mixing kernel function composed of polynomial and radial basis function (RBF) kernel functions, which uses a dynamic weight factor changing with train speeds to improve the model accuracy. The simulation results demonstrate the efficiency and accuracy of our proposed intelligent control method.

For large-scale photovoltaic module monitoring, wireless sensor network technology has been widely used, in which time synchronization is a significant factor. However, there are still outstanding issues in existing time synchronization algorithms for large-scale networking. To this end, this paper proposes a new Adaptive Energy-efficient Reference Broadcasting Synchronization method (AERBS) for wireless sensor networks, aiming to reduce the energy consumption whilst improving the synchronization precision. Moreover, the proposed method can reduce the number of synchronization information exchanges and self-adaptively determine the re-synchronization cycle. Simulation experiments demonstrate that our proposed AERBS outperforms the existing ERBS algorithm, in terms of reducing the synchronization number and the achieved reduction in energy consumption. The algorithm herein can be fully applied in the monitoring of large-scale photovoltaic module. At the same time, as a typical application of IoT (Internet of Things), it also provides a reference for distributed edge computing.

Uniform rectangular antenna array (URA) is a commonly used antenna array in multiple-input multiple-output (MIMO) system. However, the existing works rarely concentrate on URA in millimeter-wave (mmWave) line-of-sight (LOS) MIMO channels. In this paper, we conduct research on the point-to-point mmWave LOS MIMO channel based on URA. We derive the optimal antenna deployment for URA to obtain the maximum multiplexing gain in mmWave LOS MIMO channel. For the URA with fixed parameters, as the distance between transmitter and receiver increases, some singular values of the channel matrix tend to be 0. In this case, it is not necessary to use all of the transmit antennas. Therefore, according to the two kinds of channel-state information feedback schemes, we propose two adaptive transmit antenna selection (ATAS) schemes based on the principal component analysis for this mmWave LOS MIMO channel. The computational complexity for the proposed algorithm is also derived. Numerical results show that the channel capacity of this proposed ATAS system can approach or even surpass that of the full transmit antenna system with equal power allocation in mmWave LOS MIMO channel. Compared with transmit antenna selection based on greedy search and exhaustive search algorithm, when the distance is longer, the proposed algorithm can achieve the near-optimal performance with much lower computational complexity.

Citation recommendation has gained increasing attention in recent years. In practice, researchers usually prefer to cite the most topic-relevant articles. Nevertheless, how to model the implicit correlations between topics and citations is still a challenging task. In this paper, we propose a novel citation recommendation model, called TopicCite, which mines such fine-grained correlations. We extract various citation features from citation network, and integrate the learning process of feature regression with topic modeling. At the recommendation stage, we expand the folding-in process by adding the topic influence of papers that correlated with user-provided information. TopicCite can also be considered a technique for extracting topic-related citation features from manually defined citation features, which can essentially improve the granularity of pre-extracted features. In addition, the unsupervised topic model is supervised and mutually reinforced by abundant citation features in TopicCite; thus, the proposed model can also extract more reliable topic distributions from citation data, which brings a new perspective to topic discovery on linked data. The experimental results on the AAN and DBLP datasets demonstrate that our model is competitive with the state-of-the-art methods.

Due to the characteristics of uncertainty, multivariate, strong coupling, and non-linearity in the marine alkaline protease MP fermentation process, the realization of high-performance decoupling control has always been the goal. In this thesis, a nonlinear robust decoupling control method based on artificial bee colony algorithm and multiple least squares support vector machine inversion is proposed, which combines the inverse system method with support vector machine theory. First, the “gray box” dynamics model of marine alkaline protease MP fermentation process is developed based on the material balance relationship of the fermentation process, and the existence of the inverse system is analyzed. Second, the inverse model is constructed off-line using least squares support vector machine, and the inverse model is corrected on-line by artificial bee colony algorithm according to the model deviation. Finally, the inverse model is connected in series with the original system to form a combining pseudo-linear system, and according to the characteristics of the pseudo-linear system, the internal model controller is introduced to perform closed-loop control on the system. The simulation results show that the control method can realize dynamic decoupling control of marine alkaline protease fermentation process, and the system has strong parameter robustness and good anti-interference ability.

In this paper, a fast two-cycle level set contour tracking algorithm based on interactive superpixel segmentation is proposed. Then, image retrieval is carried out based on the contour tracking results. The method of contour tracking is divided into three stages: superpixel segmentation, interactive feature extraction, and foreground tracking. In the first phase, the image is preprocessed by a superpixel segmentation method. In the second stage, the feature pools are defined and established interactively according to the complexity of the image. In the last stage, based on the first two stages, a fast two-cycle level set method is used to track the contour of the target object. Finally, the results of the contour tracking are applied to an image matting of the target foreground, and then the foreground image retrieval is implemented. The results of experimental comparisons demonstrate the effectiveness of the proposed method.

In order to avoid clock skew in WSN (Wireless Sensor Networks) for large-scale photovoltaic modules monitoring, an improved time synchronization algorithm, TSP-GDM (Time Synchronization Protocol with Gaussian Delay Model) is proposed in this paper. Interdependence of local time stamps is established between network nodes according to a linear clock model. Local exchange and share of local time stamps in nodes are achieved by means of wireless transmission. An estimation method with the Gaussian Delay Model is designed to deal with the estimation problems of the node clock offset. The synchronization accuracy of the proposed method is verified with the MATLAB simulation. It is found that TSP-GDM can be applied to synchronous topology of large-scale photovoltaic modules monitoring with a higher synchronization accuracy. Compared to RBS (Reference Broadcast Synchronization), TPSN (Timing-synchronization Protocol for Sensor Networks) and RTSP (Recursive Time-Sync Protocol for WSN), its synchronization accuracy in an inner layer has been increased by 22.57μs, 15.7μs and 4.26 μs respectively.

The capacity of millimeter-wave (mmWave) line-of-sight (LOS) multiple-input multiple-output (MIMO) channel can be reduced due to the absence of rich scattering. However, the optimization of antenna array deployment can enable the maximum spatial multiplexing gain in the mmWave LOS MIMO channel, which can be applied to the long distance airborne data transmission. In this paper, we first establish the channel model of mmWave LOS MIMO based on a uniform circular antenna array (UCA) and derive the upper/lower bound of channel capacity. Aiming at achieving the maximum capacity of mmWave LOS MIMO channel, we propose the Newton-iterative method to obtain the optimal parameters of UCA. To simplify the optimization algorithm, the 2-D search can be transformed into 1-D search. Then, the analysis of bit error rate (BER) of the mmWave LOS MIMO system based on the UCA has been provided to quantify the mmWave LOS MIMO channel. Finally, numerical results demonstrate that the optimal parameters of UCA can be obtained by the Newton-iterative method and the BER of mmWave LOS MIMO communication system based on the corresponding optimized UCA can be minimized.

With the rapid growth of scientific publications, it is hard for researchers to acquire appropriate papers that meet their expectations. Recommendation system for scientific articles is an essential technology to overcome this problem. In this paper, we propose a novel low-rank and sparse matrix factorization-based paper recommendation (LSMFPRec) method for authors. The proposed method seamlessly combines low-rank and sparse matrix factorization method with fine-grained paper and author affinity matrixes that are extracted from heterogeneous scientific network. Thus, it can effectively alleviate the sparsity and cold start problems that exist in traditional matrix factorization based collaborative filtering methods. Moreover, LSMFPRec can significantly reduce the error propagated from intermediate outputs. In addition, the proposed method essentially captures the low-rank and sparse characteristics that exist in scientific rating activities; therefore, it can generate more reasonable predicted ratings for influential and uninfluential papers. The effectiveness of the proposed LSMFPRec is demonstrated by the recommendation evaluation conducted on the AAN and CiteULike data sets.

Efficient point cloud compression is essential for applications like virtual and mixed reality, autonomous driving, and cultural heritage. This paper proposes a deep learning-based inter-frame encoding scheme for dynamic point cloud geometry compression. We propose a lossy geometry compression scheme that predicts the latent representation of the current frame using the previous frame by employing a novel feature space inter-prediction network. The proposed network utilizes sparse convolutions with hierarchical multiscale 3D feature learning to encode the current frame using the previous frame. The proposed method introduces a novel predictor network for motion compensation in the feature domain to map the latent representation of the previous frame to the coordinates of the current frame to predict the current frame’s feature embedding. The framework transmits the residual of the predicted features and the actual features by compressing them using a learned probabilistic factorized entropy model. At the receiver, the decoder hierarchically reconstructs the current frame by progressively rescaling the feature embedding. The proposed framework is compared to the state-of-the-art Video-based Point Cloud Compression (V-PCC) and Geometry-based Point Cloud Compression (G-PCC) schemes standardized by the Moving Picture Experts Group (MPEG). The proposed method achieves more than 88% BD-Rate (Bjøntegaard Delta Rate) reduction against G-PCCv20 Octree, more than 56% BD-Rate savings against G-PCCv20 Trisoup, more than 62% BD-Rate reduction against V-PCC intra-frame encoding mode, and more than 52% BD-Rate savings against V-PCC P-frame-based inter-frame encoding mode using HEVC. These significant performance gains are cross-checked and verified in the MPEG working group. [ABSTRACT FROM AUTHOR]