A5100620679- Advanced Image and Video Retrieval Techniques
- Machine Learning and ELM
- Network Security and Intrusion Detection
- Infrared Target Detection Methodologies
- Video Surveillance and Tracking Methods
- Metal-Catalyzed Oxygenation Mechanisms
- Advanced Computational Techniques and Applications
- Advanced Vision and Imaging
- Porphyrin and Phthalocyanine Chemistry
- Robotics and Sensor-Based Localization
- Gaussian Processes and Bayesian Inference
- Blind Source Separation Techniques
- Advanced Clustering Algorithms Research
- Remote Sensing and Land Use
- Face and Expression Recognition
- Biomedical Text Mining and Ontologies
- Genetic and phenotypic traits in livestock
- Data Stream Mining Techniques
- Remote-Sensing Image Classification
- Guidance and Control Systems
- Advanced Image Processing Techniques
- Electrical and Bioimpedance Tomography
- Human Pose and Action Recognition
- Scientific Computing and Data Management
- Advanced Algorithms and Applications
National University of Defense Technology
2008-2023
Renmin Hospital of Wuhan University
2021
Wuhan University
2021
Sichuan University
2020
Nanjing University of Science and Technology
2015-2018
Northwestern Polytechnical University
2015
University of Science and Technology Beijing
2011-2013
Sichuan Normal University
2012
Shanghai Jiao Tong University
2007
Chongqing Technology and Business University
2003
Gaussian mixture model (GMM) clustering has been extensively studied due to its effectiveness and efficiency. Though demonstrating promising performance in various applications, it cannot effectively address the absent features among data, which is not uncommon practical applications. In this article, different from existing approaches that first impute absence then perform GMM tasks on imputed we propose integrate imputation into a unified learning procedure. Specifically, missing data...
As a representative of multiple kernel clustering (MKC), simple k-means (SimpleMKKM) is recently put forward to boosting the performance by optimally fusing group pre-specified matrices. Despite achieving significant improvement in variety applications, we find out that SimpleMKKM could indiscriminately force all sample pairs be equally aligned with same ideal similarity. result, it does not sufficiently take variation samples into consideration, leading unsatisfying performance. To address...
The synthesis of peripherally octa-substituted phthalocyanine compounds with eight strong electron-withdrawing hexylsulfonyl groups was systematically studied. Three new phthalocyanines M[Pc(SO2C6H13)8] [Pc(SO2C6H13)8 = 2,3,9,10,16,17,23,24-octakis(hexylsulfonyl)phthalocyaninate; M 2H (1), Cu (2), Zn (3)] were synthesized via direct cyclic tetramerization 4,5-di(hexylsulfonyl)phthalonitrile or through a diiminoisoindoline intermediate. Compounds 1–3 could alternatively be prepared by...
The prediction of poor ovarian response (POR) for stratified interference is a critical clinical issue that has received an increasing amount recent concern. Anthropogenic diagnostic modes remain too simple the handling actual complexity. Therefore, this study conducted extensive selection using models were derived from variety machine learning algorithms, including random forest (RF), decision trees, eXtreme Gradient Boosting (XGBoost), support vector (SVM), and artificial neural networks...
Advances in computing, networking, and multimedia technologies have led to a tremendous growth of sports video content accelerated the need analysis understanding content. Sports has been hot research area number potential applications identified. In this paper, we summarize our achievement on semantics extraction automatic editorial creation adaptation analysis. We first propose generic multi-layer multi-modal framework for Then introduce several mid-level audio/visual features which are...
Bis/tris(phthalocyaninato) europium double- and triple-decker complexes Eu [ Pc (β- SC 6 H 13 ) 8 ] 2 (1) 3 (2) = 2, 3, 9, 10, 16, 17, 23, 24-octakis(hexylthio)phthalocyaninate] have been synthesized characterized by a series of spectroscopic methods including mass, NMR, electronic absorption IR spectroscopy in addition to elemental analysis. Their molecular structures determined single crystal X-ray diffraction analysis electrochemical properties studied cyclic voltammetry.
Indoor robot localization is a challenging problem in scene recognition. Generally, appropriate image representation and multiclass classifier are the two keys to success of such task. In this paper, discriminative approach proposed meeting challenges, which composed steps: (1) spatial pyramid match Pyramid HOG (Histograms Oriented Gradient) incorporated represent an indoor place image. (2) multi-stage SVM (Support Vector Machine) utilized classify by cascade one-versus-all SVMs. The method...
Support recovery from multiple measurement vectors has been regarded as a critical aspect of compressive sensing. Most existing algorithms require the prior knowledge sparsity or noise power, which are unknown even time-varying in actual applications, to determine termination condition iterative process. Motivated by entropy concept information theory, frequency-domain (FDE)-based blind support algorithm is proposed, where FDE employed test statistic whether there sparse signal remains...
Low-dose computed tomography (LDCT) scans, which can effectively alleviate the radiation problem, will degrade imaging quality. In this paper, we propose a novel LDCT reconstruction network that unrolls iterative scheme and performs in both image manifold spaces. Because patch manifolds of medical images have low-dimensional structures, build graphs from manifolds. Then, simultaneously leverage spatial convolution to extract local pixel-level features incorporate graph analyze nonlocal...
Spatially regularised discriminative correlation filters (SRDCFs) introduce spatial regularisation weights to mitigate the boundary effects caused by circular convolution which obtains superior performance. However, is computationally expensive; this limits real‐time performance of SRDCF. This study proposes high‐speed constraint DCFs (HSCDCFs) for tracking. Using a large area sample learn CF, then, authors penalise CF coefficients. Their method formulation allows CFs efficiently mass...
In recent years, autonomous driving solutions that rely on the fusion of multiple sensors such as lidar, inertial navigation systems, and GPS positioning systems have attracted attention industry academia. Such use sets sensing to work together ensure greatest probability accuracy completeness results. These are able tackle problems difficult solve with a single sensor in short term. However, long run, design deep coupling between data strategy is not conducive real improvement perception...
We propose a transformer-based model to learn Square Word Calligraphy write English words in the format of square that resembles Chinese characters. To achieve this task, we compose dataset by collecting calligraphic characters created artist Xu Bing, and labeling position each alphabet Taking input alphabets, introduce modified relationship between predict transformation parameters for part reassemble them as character. show comparison results our predicted corresponding indicate proposed...
A neural gas network is a single-layered soft competitive network, which has many advantages for clustering analysis comparing to Kohonen's self-organizing map, K-means etc. This paper proposes splitting algorithm (SNG). By initializing neurons and finally deleting operations, the SNG can be used characterize certain class pattern effectively. We utilize construct profile of normal activities anomaly detection in security. Simulations are carried out using KDD CUP intrusion evaluation...