A5100435599- Semiconductor Lasers and Optical Devices
- Photonic and Optical Devices
- Advanced Fiber Optic Sensors
- Advanced Fiber Laser Technologies
- Optical Network Technologies
- Crime Patterns and Interventions
- Photonic Crystal and Fiber Optics
- Laser Design and Applications
- Recycled Aggregate Concrete Performance
- Municipal Solid Waste Management
- Complex Network Analysis Techniques
- Privacy-Preserving Technologies in Data
- Semiconductor Quantum Structures and Devices
- Risk and Safety Analysis
- Advanced Computational Techniques and Applications
- Evacuation and Crowd Dynamics
- Anomaly Detection Techniques and Applications
- Occupational Health and Safety Research
- Landslides and related hazards
- Advanced Neural Network Applications
- Solid State Laser Technologies
- Sustainable Supply Chain Management
- Cloud Computing and Resource Management
- Evaluation and Optimization Models
- Semantic Web and Ontologies
Sun Yat-sen University
2022-2025
Tsinghua University
2015-2024
Huaqiao University
2024
Academy of Mathematics and Systems Science
2024
Chinese Academy of Sciences
2024
Guangzhou University
2020-2024
State Key Laboratory of Geological Processes and Mineral Resources
2024
Western Michigan University
2024
Nankai University
2024
China University of Geosciences (Beijing)
2024
With the emergence of big data age, issue how to obtain valuable knowledge from a dataset efficiently and accurately has attracted increasingly attention both academia industry. This paper presents Parallel Random Forest (PRF) algorithm for on Apache Spark platform. The PRF is optimized based hybrid approach combining data-parallel task-parallel optimization. From perspective optimization, vertical data-partitioning method performed reduce communication cost effectively, data-multiplexing...
In this paper, we propose a Distributed Intelligent Video Surveillance (DIVS) system using Deep Learning (DL) algorithms and deploy it in an edge computing environment. We establish multi-layer architecture distributed DL training model for the DIVS system. The can migrate workloads from network center to edges reduce huge communication overhead provide low-latency accurate video analysis solutions. implement proposed address problems of parallel training, synchronization, workload...
A stacked ensemble model is developed for forecasting and analyzing the daily average concentrations of fine particulate matter (PM2.5) in Beijing, China. Special feature extraction procedures, including those simplification, polynomial, transformation combination, are conducted before modeling to identify potentially significant features based on an exploratory data analysis. Stability selection tree-based methods applied select important variables evaluate degrees importance. Single models...
We present RoReg, a novel point cloud registration framework that fully exploits oriented descriptors and estimated local rotations in the whole pipeline. Previous methods mainly focus on extracting rotation-invariant for but unanimously neglect orientations of descriptors. In this paper, we show are very useful pipeline, including feature description, detection, matching, transformation estimation. Consequently, design descriptor RoReg-Desc apply to estimate rotations. Such enable us...
Benefiting from convenient cycling and flexible parking locations, the Dockless Public Bicycle-sharing (DL-PBS) network becomes increasingly popular in many countries. However, redundant low-utility stations waste public urban space maintenance costs of DL-PBS vendors. In this article, we propose a Bicycle Station Dynamic Planning (BSDP) system to dynamically provide optimal bicycle station layout for network. The BSDP contains four modules: drop-off location clustering, bicycle-station...
In Vehicular Ad-hoc Networks (VANETs), privacy protection and data security during network transmission analysis have attracted attention. this paper, we apply deep learning, blockchain, fully homomorphic encryption (FHE) technologies in VANETs propose a Decentralized Privacy-preserving Deep Learning (DPDL) model. We (DVANETs) architecture, where computing tasks are decomposed from centralized cloud services to edge (EC) nodes, thereby effectively reducing communication overhead congestion...
The apical four-chamber (A4C) view in fetal echocardiography is a prenatal examination widely used for the early diagnosis of congenital heart disease (CHD). Accurate segmentation A4C key anatomical structures basis automatic measurement growth parameters and necessary diagnosis. However, due to ultrasound imaging arising from artefacts scattered noise, variability different gestational weeks, discontinuity structure boundaries, accurately segmenting organ very challenging task. To this end,...
Benefitting from large-scale training datasets and the complex network, Convolutional Neural Networks (CNNs) are widely applied in various fields with high accuracy. However, process of CNNs is very time-consuming, where large amounts samples iterative operations required to obtain high-quality weight parameters. In this paper, we focus on time-consuming propose a Bi-layered Parallel Training (BPT-CNN) architecture distributed computing environments. BPT-CNN consists two main components: (a)...
The occurrence of street crime is affected by socioeconomic and demographic characteristics also influenced streetscape conditions. Understanding how the spatial distribution associated with different features significant for establishing prevention city management strategies. Conventional data sources that quantify people on characteristics, such as questionnaires, field surveys, or manual audits, are labor-intensive, time-consuming, unable to cover a large area sufficient resolution....
In the fetal cardiac ultrasound examination, standard cycle (SCC) recognition is essential foundation for diagnosing congenital heart disease. Previous studies have mostly focused on detection of adult CCs, which may not be applicable to fetus. clinical practice, localization SCCs needs recognize end-systole (ES) and end-diastole (ED) frames accurately, ensuring that every frame in a view. Most existing methods are based key anatomical structures, irrelevant views background frames, results...