A5100357107- Advanced Algorithms and Applications
- Machine Fault Diagnosis Techniques
- Advanced Sensor and Control Systems
- Industrial Technology and Control Systems
- Fault Detection and Control Systems
- Optical measurement and interference techniques
- Advanced Computational Techniques and Applications
- Advanced Measurement and Detection Methods
- Advanced Measurement and Metrology Techniques
- Engineering Diagnostics and Reliability
- Simulation and Modeling Applications
- Optical Systems and Laser Technology
- Advanced machining processes and optimization
- Gear and Bearing Dynamics Analysis
- Human Mobility and Location-Based Analysis
- Anomaly Detection Techniques and Applications
- Chinese history and philosophy
- Power Systems and Renewable Energy
- Traffic Prediction and Management Techniques
- Power Systems and Technologies
- Indoor and Outdoor Localization Technologies
- Higher Education and Teaching Methods
- Advanced Decision-Making Techniques
- Adversarial Robustness in Machine Learning
- Advanced Machining and Optimization Techniques
National University of Defense Technology
2018-2025
Beijing Information Science & Technology University
2010-2024
Tianjin University of Technology
2012-2024
Xi'an Technological University
2008-2024
Tianjin University of Commerce
2024
Southern University of Science and Technology
2022-2024
Xuzhou Medical College
2024
Southwest Jiaotong University
2011-2024
China International Science and Technology Cooperation
2021-2024
China Academy Of Machinery Science & Technology (China)
2024
Crowd counting from unconstrained scene images is a crucial task in many real-world applications like urban surveillance and management, but it greatly challenged by the camera’s perspective that causes huge appearance variations people’s scales rotations. Conventional methods address such challenges resorting to fixed multi-scale architectures are often unable cover largely varied while ignoring rotation variations. In this paper, we propose unified neural network framework, named Deep...
Small UAV target detection plays an important role in maintaining the security of cities and citizens. targets have characteristics low-altitude flights, slow speeds, miniaturization. Taking these into account, we present a real-time algorithm called Fast-YOLOv4 based on edge computing. By adopting computing platform NVIDIA Jetson Nano, intelligent analysis can be performed video to realize fast targets. However, current iteration edge-embedded has low accuracy poor performance. To solve...
Abstract Cluster ensembles provide a framework for combining multiple base clusterings of dataset to generate stable and robust consensus clustering. There are important variants the basic cluster ensemble problem, notably including with missing values, row‐ or column‐distributed ensembles. Existing algorithms applicable only small subset these variants. In this paper, we propose Bayesian (BCE), which is mixed‐membership model learning ensembles, all primary problem. We variational...
The success of DNNs has driven the extensive applications person re-identification (ReID) into a new era. However, whether ReID inherits vulnerability remains unexplored. To examine robustness systems is rather important because insecurity may cause severe losses, e.g., criminals use adversarial perturbations to cheat CCTV systems. In this work, we current best-performing models by proposing learning-to-mis-rank formulation perturb ranking system output. As cross-dataset transferability...
An aerial Radio Environment Map (REM) characterizes the spatial distribution of Received Signal Strength (RSS) across a geographic space interest, which is crucial for optimizing wireless communication in air. Aerial REM construction can rely on Unmanned Vehicles (UAVs) to autonomously select interesting positions sampling RSS data, enhancing quality construction. However, due lack prior information about environment, it challenging UAVs determine suitable online. Additionally, achieving...
A repetitive X-band relativistic backward-wave oscillator (BWO) driven by a SINUS-881 accelerator is described. Relativistic electron beams with peak current of 5.4 kA and voltage 610 kV at repetition rate 100 Hz were generated the then guided through corrugated waveguide an axial magnetic field 3.0 T produced superconducting magnet. An collector was used to collect in order mitigate effect secondary emission electrons prevent ionization breakdown near beam dump. This BWO produces microwave...
The aim of this paper is to investigate the multiple attribute group decision making(MAGDM) problems with 2tuple linguistic assessment information, in which information about weights incompletely known, and values take form information.In order get weight vector attribute, we establish two optimization models based on basic ideal traditional TOPSIS, by can be determined.For special situations where completely unknown, some other models.By solving these models, simple exact formulas, used...
Experimental studies have shown that neuropathic pain impairs hippocampal synaptic plasticity. Here, we sought to determine the underlying mechanisms responsible for changes in painful mouse neurons. Beyond demonstrating proof-of-concept location of DExH-box helicase 9 (DHX9) nucleus, found it did exist cytoplasm and DHX9 depletion resulted structural functional at synapses hippocampus. A decrease was observed hippocampus after peripheral nerve injury; overexpression significantly alleviated...
Crowd counting from unconstrained scene images is a crucial task in many real-world applications like urban surveillance and management, but it greatly challenged by the camera's perspective that causes huge appearance variations people's scales rotations. Conventional methods address such challenges resorting to fixed multi-scale architectures are often unable cover largely varied while ignoring rotation variations. In this paper, we propose unified neural network framework, named Deep...
As a classical method to deal with nonlinear and nonstationary signals, the Hilbert–Huang transform (HHT) is widely used in various fields. In order overcome drawbacks of (such as end effects mode mixing) during process empirical decomposition (EMD), revised proposed this article. A called local linear extrapolation introduced suppress effects, combination adding high-frequency sinusoidal signal to, embedding decorrelation operator in, EMD eliminate mixing. addition, correlation coefficients...
As an emerging class of spatial trajectory data, mobile user data can be used to analyze individual or group behavioral characteristics, hobbies and interests. Besides, the information extracted from original is widely in smart cities, transportation planning, anti-terrorism maintenance. In order identify important locations target his a novel division method for preprocessing proposed, feature points are according change structural, then by clustering points, using improved density peak...
Due to numerous breakthroughs in real-world applications brought by machine intelligence, deep neural networks (DNNs) are widely employed critical applications. However, predictions of DNNs easily manipulated with imperceptible adversarial perturbations, which impedes the further deployment and may result profound security privacy implications. By incorporating samples into training data pool, is strongest principled strategy against various attacks among all sorts defense methods. Recent...
Based on the successful application of generative adversarial network (GAN) models in field image generation, this article introduces GANs into deep learning for communication systems and surveys its modulation classification. To solve difficulties feature extraction, to address low recognition accuracy existing radio signal modulation-type methods, adapt complex electromagnetic environments with high noise interference intensity, presents a model high-order digital signals. This uses Morlet...
Deep neural networks obtained by standard training have been constantly plagued adversarial examples. Although demonstrates its capability to defend against examples, unfortunately, it leads an inevitable drop in the natural generalization. To address issue, we decouple generalization and robust from joint formulate different strategies for each one. Specifically, instead of minimizing a global loss on expectation over these two errors, propose hi-expert framework called Generalist where...
The aim of this paper is to investigate the multiple attribute group decision making(MAGDM) problems with 2- tuple linguistic assessment information, in which information about weights incompletely known, and values take form information. In order get weight vector attribute, we establish two optimization models based on basic ideal traditional TOPSIS, by can be determined. For special situations where completely unknown, some other models. By solving these models, simple exact formulas,...