A5049602329- Big Data and Digital Economy
- CCD and CMOS Imaging Sensors
- Advanced Neural Network Applications
- Visual Attention and Saliency Detection
- Advanced Memory and Neural Computing
- Physical Unclonable Functions (PUFs) and Hardware Security
- Advanced Image and Video Retrieval Techniques
- Advanced Graph Neural Networks
- Smart Grid Security and Resilience
- Hydraulic and Pneumatic Systems
- Advanced Sensor and Control Systems
- Advanced Wireless Communication Techniques
- Topic Modeling
- Machine Learning in Materials Science
- Industrial Technology and Control Systems
- Neural Networks and Applications
- Machine Learning in Healthcare
- Technology and Security Systems
- Smart Grid and Power Systems
- Antenna Design and Analysis
- Advanced Algorithms and Applications
- Graph Theory and Algorithms
- Advanced MIMO Systems Optimization
Wuhan University
2022-2024
Wuhan Municipal Engineering Design & Research Institute
2022-2024
Northwest Institute of Mechanical and Electrical Engineering
2023
Beijing University of Posts and Telecommunications
2014
Beijing Microelectronics Technology Institute
2014
Differential privacy (DP) is a mathematical framework for releasing information with formal guarantees. Despite the existence of various DP procedures performing wide range statistical analysis and machine learning tasks, methods good utility are still lacking in valid inference We address this gap by introducing notion Privacy-Preserving Interval Estimation (PPIE) proposing PRivacy-loss-Efficient Consistent Inference based on poSterior quantilEs (PRECISE). PRECISE general-purpose Bayesian...
In computer vision, the joint development of algorithm and computing dimensions cannot be separated. Models algorithms are constantly evolving, while hardware designs must adapt to new or updated algorithms. Reconfigurable devices recognized as important platforms for vision applications because their reconfigurability. There two typical design approaches: customized overlay design. However, existing work is unable achieve both efficient performance scalability a wide range models. To...
Existing efforts are dedicated to designing many topologies and graph-aware strategies for the graph Transformer, which greatly improve model's representation capabilities. However, manually determining suitable Transformer architecture a specific dataset or task requires extensive expert knowledge laborious trials. This paper proposes an evolutionary search framework (EGTAS) automate construction of strong Transformers. We build comprehensive space with micro-level macro-level designs....
Security chip is widely used in security equipment of power distribution and utilization, ensuring information control systems smart grid. The paper analyzes the threat protection, such as disturbance attack, side channel attack data transmission which state cryptographic authentication are faced with terminals utilization Then this concludes that achieves goal.
Abstract In order to improve the integration degree of traditional hydraulic pneumatic elevating and banlancing machine artillery, solve centralized energy system complex pipelines all over vehicle body, combine with electro-hydrostatic actuator (EHA) technology, a double variable based on EHA technology is designed, its composition working principle are introduced, application characteristics in artillery expounded, mathematical model established. The force simulation carried out by using...
Recently, the Transformer-based models have achieved leading results in many research areas such as natural language processing and computer vision. However, since a huge computational complexities, optimization for Transformer is focus of current research. core attention module, this paper we propose software hardware co-designed which reduces units multi-head module by balancing The inference time reduced multiple modules. At same time, design corresponding structure on FPGA, accelerate...