A5032254756- Advanced SAR Imaging Techniques
- Radar Systems and Signal Processing
- Microwave Imaging and Scattering Analysis
- Target Tracking and Data Fusion in Sensor Networks
- Synthetic Aperture Radar (SAR) Applications and Techniques
- Reinforcement Learning in Robotics
- Artificial Intelligence in Games
- Advanced Measurement and Detection Methods
- Optical Systems and Laser Technology
- Advanced Algorithms and Applications
- Guidance and Control Systems
- Infrared Target Detection Methodologies
- Antenna Design and Optimization
- Geophysical Methods and Applications
- Military Defense Systems Analysis
- Digital Games and Media
- Multi-Criteria Decision Making
- Ocean Waves and Remote Sensing
- Neural Networks and Applications
- Advanced Sensor and Control Systems
- Face and Expression Recognition
- Sports Analytics and Performance
- Direction-of-Arrival Estimation Techniques
- Adversarial Robustness in Machine Learning
- Blind Source Separation Techniques
China National Institute of Standardization
2018-2025
Civil Aviation Flight University of China
2023-2024
Shanghai Jiao Tong University
2020-2024
Henan University of Economic and Law
2024
Shenyang University of Chemical Technology
2024
National University of Defense Technology
2014-2023
Air Force Engineering University
2018-2023
Tencent (China)
2020-2023
NARI Group (China)
2023
Microsoft Research Asia (China)
2023
This paper proposes a synthetic aperture radar (SAR) automatic target recognition approach based on global scattering center model. The model is established offline using range profiles at multiple viewing angles, so the original data amount much less than that required for establishing SAR image templates. Scattering features different poses can be conveniently predicted by this Moreover, modified to predict various configurations. For classified, regional in levels are extracted...
We study the reinforcement learning problem of complex action control in Multi-player Online Battle Arena (MOBA) 1v1 games. This involves far more complicated state and spaces than those traditional games, such as Go Atari series, which makes it very difficult to search any policies with human-level performance. In this paper, we present a deep framework tackle from perspectives both system algorithm. Our is low coupling high scalability, enables efficient explorations at large scale....
MOBA games, e.g., Honor of Kings, League Legends, and Dota 2, pose grand challenges to AI systems such as multi-agent, enormous state-action space, complex action control, etc. Developing for playing games has raised much attention accordingly. However, existing work falls short in handling the raw game complexity caused by explosion agent combinations, i.e., lineups, when expanding hero pool case that OpenAI's limits play a only 17 heroes. As result, full without restrictions are far from...
Structural magnetic resonance imaging (sMRI), especially longitudinal sMRI, is often used to monitor and capture disease progression during the clinical diagnosis of Alzheimer's Disease (AD). However, current methods neglect AD's progressive nature have mostly relied on a single image for recognizing AD. In this paper, we consider problem leveraging MRIs subject AD classification. To address challenges missing data, data demand, subtle changes over time in learning 3D MRIs, propose novel...
In this letter, a 3-D range migration algorithm for multiple-input multiple-output synthetic aperture radar imaging is proposed. The accurate expression of the signal spectrum derived by utilizing spherical wave decomposition. This method compensates curvature wavefront in wavenumber domain and achieves Fourier transform reflectivity map through dimension-reducing accumulation operation. Its fast transform-based scheme provides high efficiency. addition, does not take plane approximation can...
The scale of ground-to-air confrontation task assignments is large and needs to deal with many concurrent random events. Aiming at the problems where existing assignment methods are applied confrontation, there low efficiency in dealing complex tasks, interactive conflicts multiagent systems. This study proposes a architecture based on one-general agent multiple narrow agents (OGMN) reduce conflicts. Considering slow speed traditional dynamic algorithms, this paper proximal policy...
Traffic crash prediction (TCP) is a fundamental problem for intelligent transportation systems in smart cities. Improving the accuracy of traffic important road safety and effective management. Owing to recent advances artificial neural networks, several new deep-learning models have been proposed TCP. However, these works mainly focus on accidents regions, which are typically pre-determined using grid map. We argue that TCP roads, especially crashes at or near intersections account more...
Aggregation of intuitionistic fuzzy information is a hot topic in Atanassov's set theory, which has attracted much interest from researchers recent years. In this paper, series new aggregation operators and weighted averaging are proposed for aggregating information. First, some basic laws operations on values presented together with their properties. Then, we propose arithmetic operator geometric to aggregate Inspired by the idea ordered hybrid averaging, further develop operator, (IFHWAA)...
We present JueWu-SL, the first supervised-learning-based artificial intelligence (AI) program that achieves human-level performance in playing multiplayer online battle arena (MOBA) games. Unlike prior attempts, we integrate macro-strategy and micromanagement of MOBA-game-playing into neural networks a supervised end-to-end manner. Tested on Honor Kings, most popular MOBA at present, our AI performs competitively level High King players standard 5v5
The data-driven convolutional neural networks (CNNs) have achieved great progress in Synthetic Aperture Radar automatic target recognition (SAR-ATR) after being trained a large scale of labeled samples.However, the insufficiency SAR data always leads to over-fitting, causing significant performance degradation.To solve mentioned problem, semi-supervised transfer learning method based on generative adversarial (GANs) is presented present paper.The discriminator GAN with an encoder and...
Inverse synthetic aperture radar (ISAR) can form two‐dimensional (2D) electromagnetic images of a target, but it cannot provide the third dimensional information about target. Conventional 3D turntable ISAR imaging requires data collection over densely azimuth‐elevation samples, which needs large amount storage. In this study, an effective algorithm for model based on compressive sensing is proposed, exploits sparsity in image domain to achieve reconstruction by using limited number...
A methodology to reconstruct the 3-D scattering center model from data with wide azimuthal aperture at a single elevation, such as those collected in turntable or circular synthetic radar configurations, is proposed this paper. The divided into overlapped subapertures, and 2-D centers are extracted each subaperture. These local rotated mapped ground plane target coordinate system, where they associated according their location amplitude consistency. Three-dimensional position of scatterer...
Hero drafting is essential in multiplayer online battle arena (MOBA) game playing as it builds the team of each side and directly affects match outcome. State-of-the-art methods fail to consider: 1) efficiency when hero pool expanded; 2) multiround nature a MOBA 5v5 series, i.e., two teams play best-of- <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$N$</tex-math></inline-formula> same only allowed be...
Existing missile defense target threat assessment methods ignore the timing and battlefield changes, leading to low accuracy. In order overcome this problem, a dynamic multi-time fusion method is proposed. method, new interval valued intuitionistic fuzzy weighted averaging operator proposed effectively aggregate multi-source uncertain information; an interval-valued entropy based on cosine function (IVIFECF) designed determine attribute weight; improved number distance measurement model...
An effective ensemble should consist of a set networks that are both accurate and diverse. We propose novel clustering-based selective algorithm for constructing neural network ensemble, where clustering technology is used to classify trained according similarity optimally select the most individual from each cluster make up ensemble. Empirical studies on regression four typical datasets showed this approach yields significantly smaller achieving better performance than other traditional...
The ultimate goal of military intelligence is to equip the command and control (C2) system with decision-making art excellent human commanders be more agile stable than beings. Intelligent commander Alpha C2 solves dynamic problem in complex scenarios air defense operations using a deep reinforcement learning framework. Unlike traditional systems that rely on expert rules models, interacts digital battlefields close real world generates data. By integrating states multiple parties as input,...
Deep convolutional neural networks (CNNs) have made a breakthrough on supervised SAR images classification. However, imaging is considerably affected by the frequency band. That means network trained image set of one band not suitable for classification another images. As manually labeling training samples each always time-consuming, we propose an unsupervised multi-level domain adaptation method based adversarial learning to solve problem multi-band First, train discriminative CNN using...
Learning rational behaviors in open-world games like Minecraft remains to be challenging for Reinforcement (RL) research due the compound challenge of partial observability, high-dimensional visual perception and delayed reward. To address this, we propose JueWu-MC, a sample-efficient hierarchical RL approach equipped with representation learning imitation deal exploration. Specifically, our includes two levels hierarchy, where high-level controller learns policy control over options...
A novel machine learning method named extended support vector data description with negative examples (ESVDD-neg) is developed to classify the fast Fourier transform-magnitude feature of complex high-resolution range profile (HRRP), motivated by problem radar automatic target recognition. The proposed not only inherits close non-linear boundary advantage model but also incorporates a new paradigm using privileged information into model. It leads appealing application no assumptions regarding...
In near-range sparse multiple-input multiple-output (MIMO) array imaging, since grating lobes spread wide and overlap with each other, the zeros of transmitting receiving cannot cancel other out. As a result, there are residual in imaging results. This article proposes lobe suppression method based on zero migration multiapodization. The is equivalent to broadening array. Therefore, completely suppressed without increasing redundancy To realize migration, novel named generalized matched...