arXiv subjects

Wei Xiang

Publications and source records attributed to Wei Xiang.

At least 109 records · Page 6Linked to original sources

arXiv2022

DDU-Net: Dual-Decoder-U-Net for Road Extraction Using High-Resolution Remote Sensing Images

Extracting roads from high-resolution remote sensing images (HRSIs) is vital in a wide variety of applications, such as autonomous driving, path planning, and road navigation. Due to the long and thin shape as well as the shades induced by vegetation and buildings, small-sized roads are more difficult to discern. In order to improve the reliability and accuracy of small-sized road extraction when roads of multiple sizes coexist in an HRSI, an enhanced deep neural network model termed Dual-Decoder-U-Net (DDU-Net) is proposed in this paper. Motivated by the U-Net model, a small decoder is added to form a dual-decoder structure for more detailed features. In addition, we introduce the dilated convolution attention module (DCAM) between the encoder and decoders to increase the receptive field as well as to distill multi-scale features through cascading dilated convolution and global average pooling. The convolutional block attention module (CBAM) is also embedded in the parallel dilated convolution and pooling branches to capture more attention-aware features. Extensive experiments are conducted on the Massachusetts Roads dataset with experimental results showing that the proposed model outperforms the state-of-the-art DenseUNet, DeepLabv3+ and D-LinkNet by 6.5%, 3.3%, and 2.1% in the mean Intersection over Union (mIoU), and by 4%, 4.8%, and 3.1% in the F1 score, respectively. Both ablation and heatmap analyses are presented to validate the effectiveness of the proposed model.

cs.CV↗

arXiv2021

Stability of transonic contact discontinuity for two-dimensional steady compressible Euler flows in a finitely long nozzle

We consider the stability of transonic contact discontinuity for the two-dimensional steady compressible Euler flows in a finitely long nozzle. This is the first work on the mixed-type problem of transonic flows across a contact discontinuity as a free boundary in nozzles. We start with the Euler-Lagrangian transformation to straighten the contact discontinuity in the new coordinates. However, the upper nozzle wall in the subsonic region depending on the mass flux becomes a free boundary after the transformation. Then we develop new ideas and techniques to solve the free-boundary problem in three steps: (1) we fix the free boundary and generate a new iteration scheme to solve the corresponding fixed boundary value problem of the hyperbolic-elliptic mixed type by building some powerful estimates for both the first-order hyperbolic equation and a second-order nonlinear elliptic equation in a Lipschitz domain; (2) we update the new free boundary by constructing a mapping that has a fixed point; (3) we establish via the inverse Lagrangian coordinates transformation that the original free interface problem admits a unique piecewise smooth transonic solution near the background state, which consists of a smooth subsonic flow and a smooth supersonic flow with a contact discontinuity.

math.AP↗

arXiv2021

TE-YOLOF: Tiny and efficient YOLOF for blood cell detection

Blood cell detection in microscopic images is an essential branch of medical image processing research. Since disease detection based on manual checking of blood cells is time-consuming and full of errors, testing of blood cells using object detectors with Deep Convolutional Neural Network can be regarded as a feasible solution. In this work, an object detector based on YOLOF has been proposed to detect blood cell objects such as red blood cells, white blood cells and platelets. This object detector is called TE-YOLOF, Tiny and Efficient YOLOF, and it is a One-Stage detector using dilated encoder to extract information from single-level feature maps. For increasing efficiency and flexibility, the EfficientNet Convolutional Neural Network is utilized as the backbone for the proposed object detector. Furthermore, the Depthwise Separable Convolution is applied to enhance the performance and minimize the parameters of the network. In addition, the Mish activation function is employed to increase the precision. Extensive experiments on the BCCD dataset prove the effectiveness of the proposed model, which is more efficient than other existing studies for blood cell detection.

cs.CV↗

arXiv2021

Persistence of the steady planar normal shock structure in 3-D unsteady potential flows

This paper concerns the dynamic stability of the steady 3-D wave structure of a planar normal shock front intersecting perpendicularly to a planar solid wall for unsteady potential flows. The stability problem can be formulated as a free boundary problem of a quasi-linear hyperbolic equation of second order in a dihedral-space domain between the shock front and the solid wall. The key difficulty is brought by the edge singularity of the space domain, the intersection curve between the shock front and the solid wall. Different from the 2-D case, for which the singular part of the boundary is only a point, it is a curve for the 3-D case in this paper. This difference brings new difficulties to the mathematical analysis of the stability problem. A modified partial hodograph transformation is introduced such that the extension technique developed for the 2-D case can be employed to establish the well-posed theory for the initial-boundary value problem of the linearized hyperbolic equation of second order in a dihedral-space domain. Moreover, the extension technique is improved in this paper such that loss of regularity in the a priori estimates on the shock front does not occur. Thus the classical nonlinear iteration scheme can be constructed to prove the existence of the solution to the stability problem, which shows the dynamic stability of the steady planar normal shock without applying the Nash-Moser iteration method.

math.AP↗

arXiv2021

Stability of Attached Transonic Shocks in Steady Potential Flow past Three-Dimensional Wedges

We develop a new approach and employ it to establish the global existence and nonlinear structural stability of attached weak transonic shocks in steady potential flow past three-dimensional wedges; in particular, the restriction that the perturbation is away from the wedge edge in the previous results is removed. One of the key ingredients is to identify a "good" direction of the boundary operator of a boundary condition of the shock along the wedge edge, based on the non-obliqueness of the boundary condition for the weak shock on the edge. With the identification of this direction, an additional boundary condition on the wedge edge can be assigned to make sure that the shock is attached on the edge and linearly stable under small perturbation. Based on the linear stability, we introduce an iteration scheme and prove that there exists a unique fixed point of the iteration scheme, which leads to the global existence and nonlinear structural stability of the attached weak transonic shock. This approach is based on neither the hodograph transformation nor the spectrum analysis, and should be useful for other problems with similar difficulties.

math.AP↗

arXiv2021

Convergence Rate of Hypersonic Similarity for Steady Potential Flows Over Two-Dimensional Lipschitz Wedge

This paper is devoted to establishing the convergence rate of the hypersonic similarity for the inviscid steady irrotational Euler flow over a two-dimensional Lipschitz slender wedge in $BV\cap L^1$ space. The rate we established is the same as the one predicted by Newtonian-Busemann law (see (3.29) in \cite[Page 67]{anderson} for more details)as the incoming Mach number $\textrm{M}_{\infty}\rightarrow\infty$ for a fixed hypersonic similarity parameter $K$. The hypersonic similarity, which is also called the Mach-number independence principle, is equivalent to the following Van Dyke's similarity theory: For a given hypersonic similarity parameter $K$, when the Mach number of the flow is sufficiently large, the governing equations after the scaling are approximated by a simpler equation, that is called the hypersonic small-disturbance equation. To achieve the convergence rate, we approximate the curved boundary by piecewisely straight lines and find a new Lipschitz continuous map $\mathcal{P}_{h}$ such that the trajectory can be obtained by piecing together the Riemann solutions near the approximated boundary. Next, we derive the $L^1$ difference estimates between the approximate solutions $U^{(τ)}_{h,ν}(x,\cdot)$ to the initial-boundary value problem for the scaled equations and the trajectories $\mathcal{P}_{h}(x,0)(U^ν_{0})$ by piecing together all the Riemann solvers. Then, by the uniqueness and the compactness of $\mathcal{P}_{h}$ and $U^{(τ)}_{h,ν}$, we can further establish the $L^1$ estimates of order $τ^2$ between the solutions to the initial-boundary value problem for the scaled equations and the solutions to the initial-boundary value problem for the hypersonic small-disturbance equations, if the total variations of the initial data and the tangential derivative of the boundary are sufficiently small.

math.AP↗

arXiv2021

AGSFCOS: Based on attention mechanism and Scale-Equalizing pyramid network of object detection

Recently, the anchor-free object detection model has shown great potential for accuracy and speed to exceed anchor-based object detection. Therefore, two issues are mainly studied in this article: (1) How to let the backbone network in the anchor-free object detection model learn feature extraction? (2) How to make better use of the feature pyramid network? In order to solve the above problems, Experiments show that our model has a certain improvement in accuracy compared with the current popular detection models on the COCO dataset, the designed attention mechanism module can capture contextual information well, improve detection accuracy, and use sepc network to help balance abstract and detailed information, and reduce the problem of semantic gap in the feature pyramid network. Whether it is anchor-based network model YOLOv3, Faster RCNN, or anchor-free network model Foveabox, FSAF, FCOS. Our optimal model can get 39.5% COCO AP under the background of ResNet50.

cs.CV↗

arXiv2021

DPN-SENet:A self-attention mechanism neural network for detection and diagnosis of COVID-19 from chest x-ray images

Background and Objective: The new type of coronavirus is also called COVID-19. It began to spread at the end of 2019 and has now spread across the world. Until October 2020, It has infected around 37 million people and claimed about 1 million lives. We propose a deep learning model that can help radiologists and clinicians use chest X-rays to diagnose COVID-19 cases and show the diagnostic features of pneumonia. Methods: The approach in this study is: 1) we propose a data enhancement method to increase the diversity of the data set, thereby improving the generalization performance of the model. 2) Our deep convolution neural network model DPN-SE adds a self-attention mechanism to the DPN network. The addition of a self-attention mechanism has greatly improved the performance of the network. 3) Use the Lime interpretable library to mark the feature regions on the X-ray medical image that helps doctors more quickly diagnose COVID-19 in people. Results: Under the same network model, the data with and without data enhancement is put into the model for training respectively. At last, comparing two experimental results: among the 10 network models with different structures, 7 network models have improved their effects after using data enhancement, with an average improvement of 1% in recognition accuracy. We propose that the accuracy and recall rates of the DPN-SE network are 93% and 98% of cases (COVID vs. pneumonia bacteria vs. viral pneumonia vs. normal). Compared with the original DPN, the respective accuracy is improved by 2%. Conclusion: The data augmentation method we used has achieved effective results on a small amount of data set, showing that a reasonable data augmentation method can improve the recognition accuracy without changing the sample size and model structure. Overall, the proposed method and model can effectively become a very useful tool for clinical radiologists.

eess.IV↗

arXiv2021

Denoising Higher-order Moments for Blind Digital Modulation Identification in Multiple-antenna Systems

The paper proposes a new technique that substantially improves blind digital modulation identification (DMI) algorithms that are based on higher-order statistics (HOS). The proposed technique takes advantage of noise power estimation to make an offset on higher-order moments (HOM), thus getting an estimate of noise-free HOM. When tested for multiple-antenna systems, the proposed method outperforms other DMI algorithms, in terms of identification accuracy, that are based only on cumulants or do not consider HOM denoising, even for a receiver with impairments. The improvement is achieved with the same order of complexity of the common HOS-based DMI algorithms in the same context.

eess.SP↗

arXiv2021

Towards Memristive Deep Learning Systems for Real-time Mobile Epileptic Seizure Prediction

The unpredictability of seizures continues to distress many people with drug-resistant epilepsy. On account of recent technological advances, considerable efforts have been made using different hardware technologies to realize smart devices for the real-time detection and prediction of seizures. In this paper, we investigate the feasibility of using Memristive Deep Learning Systems (MDLSs) to perform real-time epileptic seizure prediction on the edge. Using the MemTorch simulation framework and the Children's Hospital Boston (CHB)-Massachusetts Institute of Technology (MIT) dataset we determine the performance of various simulated MDLS configurations. An average sensitivity of 77.4% and a Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.85 are reported for the optimal configuration that can process Electroencephalogram (EEG) spectrograms with 7,680 samples in 1.408ms while consuming 0.0133W and occupying an area of 0.1269mm$^2$ in a 65nm Complementary Metal-Oxide-Semiconductor (CMOS) process.

cs.ET↗

arXiv2021

Performance Analysis for Cache-enabled Cellular Networks with Cooperative Transmission

The large amount of deployed smart devices put tremendous traffic pressure on networks. Caching at the edge has been widely studied as a promising technique to solve this problem. To further improve the successful transmission probability (STP) of cache-enabled cellular networks (CEN), we combine the cooperative transmission technique with CEN and propose a novel transmission scheme. Local channel state information (CSI) is introduced at each cooperative base station (BS) to enhance the strength of the signal received by the user. A tight approximation for the STP of this scheme is derived using tools from stochastic geometry. The optimal content placement strategy of this scheme is obtained using a numerical method to maximize the STP. Simulation results demonstrate the optimal strategy achieves significant gains in STP over several comparative baselines with the proposed scheme.

cs.IT↗

arXiv2021

Probabilistic Placement Optimization for Non-coherent and Coherent Joint Transmission in Cache-Enabled Cellular Networks

How to design proper content placement strategies is one of the major areas of interest in cache-enabled cellular networks. In this paper, we study the probabilistic content placement optimization of base station (BS) caching with cooperative transmission in the downlink of cellular networks. With placement probability vector being the design parameter, non-coherent joint transmission (NC-JT) and coherent joint transmission (C-JT) schemes are investigated according to whether channel state information (CSI) is available. Using stochastic geometry, we derive an integral expression for the successful transmission probability (STP) in NC-JT scheme, and present an upper bound and a tight approximation for the STP of the C-JT scheme. Next, we maximize the STP in NC-JT and the approximation of STP in C-JT by optimizing the placement probability vector, respectively. An algorithm is proposed and applied to both optimization problems. By utilizing some properties of the STP, we obtain globally optimal solutions in certain cases. Moreover, locally optimal solutions in general cases are obtained by using the interior point method. Finally, numerical results show the optimized placement strategy achieves significant gains in STP over several comparative baselines both in NC-JT and C-JT. The optimal STP in C-JT outperforms the one in NC-JT, indicating the benefits of knowing CSI in cooperative transmission.

cs.IT↗

arXiv2021

Internet of Underwater Things and Big Marine Data Analytics -- A Comprehensive Survey

The Internet of Underwater Things (IoUT) is an emerging communication ecosystem developed for connecting underwater objects in maritime and underwater environments. The IoUT technology is intricately linked with intelligent boats and ships, smart shores and oceans, automatic marine transportations, positioning and navigation, underwater exploration, disaster prediction and prevention, as well as with intelligent monitoring and security. The IoUT has an influence at various scales ranging from a small scientific observatory, to a midsized harbor, and to covering global oceanic trade. The network architecture of IoUT is intrinsically heterogeneous and should be sufficiently resilient to operate in harsh environments. This creates major challenges in terms of underwater communications, whilst relying on limited energy resources. Additionally, the volume, velocity, and variety of data produced by sensors, hydrophones, and cameras in IoUT is enormous, giving rise to the concept of Big Marine Data (BMD), which has its own processing challenges. Hence, conventional data processing techniques will falter, and bespoke Machine Learning (ML) solutions have to be employed for automatically learning the specific BMD behavior and features facilitating knowledge extraction and decision support. The motivation of this paper is to comprehensively survey the IoUT, BMD, and their synthesis. It also aims for exploring the nexus of BMD with ML. We set out from underwater data collection and then discuss the family of IoUT data communication techniques with an emphasis on the state-of-the-art research challenges. We then review the suite of ML solutions suitable for BMD handling and analytics. We treat the subject deductively from an educational perspective, critically appraising the material surveyed.

cs.NI↗

arXiv2020

Dynamic Energy Dispatch Based on Deep Reinforcement Learning in IoT-Driven Smart Isolated Microgrids

Microgrids (MGs) are small, local power grids that can operate independently from the larger utility grid. Combined with the Internet of Things (IoT), a smart MG can leverage the sensory data and machine learning techniques for intelligent energy management. This paper focuses on deep reinforcement learning (DRL)-based energy dispatch for IoT-driven smart isolated MGs with diesel generators (DGs), photovoltaic (PV) panels, and a battery. A finite-horizon Partial Observable Markov Decision Process (POMDP) model is formulated and solved by learning from historical data to capture the uncertainty in future electricity consumption and renewable power generation. In order to deal with the instability problem of DRL algorithms and unique characteristics of finite-horizon models, two novel DRL algorithms, namely, finite-horizon deep deterministic policy gradient (FH-DDPG) and finite-horizon recurrent deterministic policy gradient (FH-RDPG), are proposed to derive energy dispatch policies with and without fully observable state information. A case study using real isolated MG data is performed, where the performance of the proposed algorithms are compared with the other baseline DRL and non-DRL algorithms. Moreover, the impact of uncertainties on MG performance is decoupled into two levels and evaluated respectively.

cs.LG↗

arXiv2020

Performance Analysis of Low-Interference N-Continuous OFDM

The low-interference N-continuous orthogonal frequency division multiplexing (NC-OFDM) system [25], [26] is investigated in terms of power spectrum density (PSD) and bit error rate (BER), to prove and quantify its advantages over traditional NC-OFDM. The PSD and BER performances of the low-interference scheme are analyzed and compared under the parameters of the highest derivative order (HDO) and the length of the smooth signal. In the context of PSD, different from one discontinuous point per NC-OFDM symbol in [25], the sidelobe suppression performance is evaluated upon considering two discontinuous points due to the finite continuity of the smooth signal and its higher-order derivatives. It was shown that with an increased HDO and an increased length of the smooth signal, a more rapid sidelobe decaying is achieved, for the significant continuity improvement of the OFDM signal and its higher-order derivatives. However, our PSD analysis also shows that if the length of the smooth signal is set inappropriately, the performance may be degraded, even if the HDO is large. Furthermore, it was shown in the error performance analysis that under the assumptions of perfect and imperfect synchronization, the low-interference scheme incurs small BER performance degradation for a short length of the smooth signal or a small HDO as opposed to conventional NC-OFDM. Based on analysis and simulation results, the trade-offs between sidelobe suppression and BER are studied with the above two parameters.

cs.IT↗

arXiv2020

FASTSWARM: A Data-driven FrAmework for Real-time Flying InSecT SWARM Simulation

Insect swarms are common phenomena in nature and therefore have been actively pursued in computer animation. Realistic insect swarm simulation is difficult due to two challenges: high-fidelity behaviors and large scales, which make the simulation practice subject to laborious manual work and excessive trial-and-error processes. To address both challenges, we present a novel data-driven framework, FASTSWARM, to model complex behaviors of flying insects based on real-world data and simulate plausible animations of flying insect swarms. FASTSWARM has a linear time complexity and achieves real-time performance for large swarms. The high-fidelity behavior model of FASTSWARM explicitly takes into consideration the most common behaviors of flying insects, including the interactions among insects such as repulsion and attraction, the self-propelled behaviors such as target following and obstacle avoidance, and other characteristics such as the random movements. To achieve scalability, an energy minimization problem is formed with different behaviors modelled as energy terms, where the minimizer is the desired behavior. The minimizer is computed from the real-world data, which ensures the plausibility of the simulation results. Extensive simulation results and evaluations show that FASTSWARM is versatile in simulating various swarm behaviors, high fidelity measured by various metrics, easily controllable in inducing user controls and highly scalable.

cs.GR↗

arXiv2020

Detached shock past a blunt body

In $\R^2$, a symmetric blunt body $W_b$ is fixed by smoothing out the tip of a symmetric wedge $W_0$ with the half-wedge angle $θ_w\in (0, \fracπ{2})$. We first show that if a horizontal supersonic flow of uniform state moves toward $W_0$ with a Mach number $M_{\infty}>1$ sufficiently large, %depending on $θ_w$, then there exist two shock solutions, {\emph{a weak shock solution and a strong shock solution}}, with the shocks being straight and attached to the tip of the wedge $W_0$. Such shock solutions are given by a shock polar analysis, and they satisfy entropy conditions. The main goal of this work is to construct a detached shock solution of the steady Euler system for inviscid compressible irrotational flow in $\R^2\setminus W_b$. In particular, we seek a shock solution with the far-field state being the strong shock solution obtained from the shock polar analysis. Furthermore, we prove that the detached shock forms a convex curve around the blunt body $W_b$ if the Mach number of the incoming supersonic flow is sufficiently large, and if the boundary of $W_b$ is convex.

math.AP↗

arXiv2020

Convexity of Self-Similar Transonic Shocks and Free Boundaries for the Euler Equations for Potential Flow

We are concerned with geometric properties of transonic shocks as free boundaries in two-dimensional self-similar coordinates for compressible fluid flows, which are not only important for the understanding of geometric structure and stability of fluid motions in continuum mechanics but also fundamental in the mathematical theory of multidimensional conservation laws. A transonic shock for the Euler equations for self-similar potential flow separates elliptic (subsonic) and hyperbolic (supersonic) phases of the self-similar solution of the corresponding nonlinear partial differential equation in a domain under consideration, in which the location of the transonic shock is apriori unknown. We first develop a general framework under which self-similar transonic shocks, as free boundaries, are proved to be uniformly convex, and then apply this framework to prove the uniform convexity of transonic shocks in the two longstanding fundamental shock problems -- the shock reflection-diffraction by wedges and the Prandtl-Meyer reflection for supersonic flows past solid ramps. To achieve this, our approach is to exploit underlying nonlocal properties of the solution and the free boundary for the potential flow equation.

math.AP↗