A5040071999- Advanced Neural Network Applications
- Multimodal Machine Learning Applications
- Vehicle License Plate Recognition
- Video Surveillance and Tracking Methods
- Renin-Angiotensin System Studies
- Advanced Image and Video Retrieval Techniques
- Simulation Techniques and Applications
- Infrastructure Maintenance and Monitoring
- Digital Media Forensic Detection
- Topic Modeling
- Generative Adversarial Networks and Image Synthesis
- Machine Learning and Data Classification
- Tensor decomposition and applications
- Face and Expression Recognition
- Geographic Information Systems Studies
- Image Processing and 3D Reconstruction
- Adaptive Dynamic Programming Control
- Advanced Computing and Algorithms
- Intracerebral and Subarachnoid Hemorrhage Research
- Advanced Technologies in Various Fields
- AI and Multimedia in Education
- Recommender Systems and Techniques
- Ferroptosis and cancer prognosis
- Domain Adaptation and Few-Shot Learning
- Handwritten Text Recognition Techniques
Fuzhou University
2024-2025
Daping Hospital
2024
Army Medical University
2024
Meizu (China)
2024
Zhejiang University of Science and Technology
2021-2023
Background: Hemorrhagic shock was a leading cause of death worldwide, with myocardial injury being primary affected organ. As commonly used solutions in fluid resuscitation, acetated Ringer's (AR) and Lactate solution (LR) were far from perfect for their adverse reactions such as lactic acidosis electrolyte imbalances. In previous studies, TPP@PAMAM-MR (TPP-MR), novel nanocrystal resuscitation has been found to protect against septic rats. However, its role rats hemorrhagic underlying...
Multi-view clustering has attracted significant attention in recent years because it can leverage the consistent and complementary information of multiple views to improve performance. However, effectively fuse balance are common challenges faced by multi-view clustering. Most existing fusion works focus on weighted-sum concatenating fusion, which unable fully underlying information, not consider balancing views. To this end, we propose Cross-view Fusion for Clustering (CFMVC). Specifically,...
An improved Ghost-YOLOv5s detection algorithm is proposed in this paper to solve the problems of high computational load and undesirable recognition rate traditional methods pavement diseases. Ghost modules C3Ghost are introduced into YOLOv5s network reduce FLOPs (floating-point operations) feature channel fusion process. Mosaic data augmentation also added improve expression performance. A public road disease dataset reconstructed verify performance method. The model trained deployed NVIDIA...
Multi-view clustering aims to improve the performance by leveraging information from multiple views. Most existing works assume that all views are complete. However, samples in real-world scenarios cannot be always observed views, leading challenging problem of Incomplete Multi-View Clustering (IMVC). Although some attempts made recently, they still suffer following two limitations: (1) usually adopt shallow models, which unable sufficiently explore consistency and complementary views; (2)...
Pre-trained vision-language (V-L) models such as CLIP have shown excellent performance in many downstream cross-modal tasks. However, most of them are only applicable to the English context. Subsequent research has focused on this problem and proposed improved models, CN-CLIP AltCLIP, facilitate their applicability Chinese even other languages. Nevertheless, these suffer from high latency a large memory footprint inference, which limits further deployment resource-constrained edge devices....
YOLO (You Only Look Once), as a target detection algorithm with good speed and precision, is widely used in the industry. In process of driving, vehicle image captured by driving camera detected it extracts license plate front part vehicle. Compared network structure YOLOv3-tiny algorithm, acquisition method anchor box improved combining Birch algorithm. order to improve real-time performance, original two-scale added multi-scale prediction three-scale ensure its accuracy. Finally,...
Abstract With the rapid development of artificial intelligence technology, commercial robots have gradually entered our daily lives. In order to promote product dissemination, shopping guide are a new service options commerce platforms that use tag recommendation systems identify users' intentions. A large number applications combine user historical tagging information with multi‐round dialogue ability help users efficiently search for and retrieve products interest. Recently, tensor...
Object counting is a challenging task with broad application prospects in security surveillance, traffic management, and disease diagnosis. Existing object methods face tri-fold challenge: achieving superior performance, maintaining high generalizability, minimizing annotation costs. We develop novel training-free class-agnostic counter, TFCounter, which prompt-context-aware via the cascade of essential elements large-scale foundation models. This approach employs an iterative framework dual...
Traditionally, firms have offered coupons to customer groups at predetermined discount rates. However, advancements in machine learning and the availability of abundant data now enable platforms provide real-time customized individuals. In this study, we partner with Meituan, a leading shopping platform, develop real-time, end-to-end coupon allocation system that is fast effective stimulating demand while adhering marketing budgets when faced uncertain traffic from diverse base. Leveraging...
Asynchronous advantage actor-critic (A3C) algorithm is a commonly used policy optimization in reinforcement learning, which asynchronous parallel interactive sampling and training, multi-step reward estimation method for computing weights. In order to address the problem of low efficiency insufficient convergence caused by traditional heuristic exploration A3C an improved proposed this paper. algorithm, noise network function, updates tensor explicit way constructed train agent. Generalised...