A5104267630- Advanced Neural Network Applications
- Industrial Technology and Control Systems
- Advanced Computational Techniques and Applications
- Algorithms and Data Compression
- Electric Motor Design and Analysis
- Remote Sensing and LiDAR Applications
- Engineering and Materials Science Studies
- Advanced Memory and Neural Computing
- Advanced Algorithms and Applications
- Web Data Mining and Analysis
- Magnetic Bearings and Levitation Dynamics
- Metal Forming Simulation Techniques
- VLSI and Analog Circuit Testing
- Cognitive Computing and Networks
- Network Security and Intrusion Detection
- Inertial Sensor and Navigation
- Metallurgy and Material Forming
- Embedded Systems Design Techniques
- Spam and Phishing Detection
- Network Packet Processing and Optimization
- Advanced Sensor and Control Systems
- Sensorless Control of Electric Motors
- Advanced Image and Video Retrieval Techniques
- Peer-to-Peer Network Technologies
- Traffic Prediction and Management Techniques
Southeast University
2023-2025
Beijing Institute of Technology
2006-2007
Fast sorting of large-scale data is an essential task for centers. In previous works, the existing computational model kernel still results in lower bandwidth utilization on external memory bus. And execution merge operations sort circuit FPGAs depends control commands from host CPU. this case, not fully offloaded to hardware layer acceleration, resulting a performance loss. We design on-chip controller efficiently command process. The proposed has ability schedule multiple computing kernels...
Abstract Transformers have been widely used in various computer vision applications. Compared to traditional convolutional neural networks (CNNs), transformer's inference includes plenty of non‐linear operations, such as softmax and Gaussian error linear units (GELU). As the scale transformers grows, an efficient hardware implementation these operations is significant. However, current works network accelerators focus on CNN less attention paid transformer. In addition, most FPGA‐based or...
Fast regular expression matching is an essential task for deep packet inspection. In previous works, the engine on FPGA struggled to achieve ideal balance between resource consumption and throughput. Speculation enumerative computation exploits statistical properties of deterministic finite automata, allowing more efficient pattern matching. Existing related designs mostly revolve around vector instructions multiple processors/cores or SIMD instruction sets, with a lack implementation...
To improve the predication accuracy for web traffic, a method was proposed based on integration of wavelet analysis and neural network. The traffic time series, which is nonlinear non-stationary, decomposed and, then, reconstructed into several branches by method. These were predicted networks respectively final value combination these results. Theoretical experiment results show that can decompose original series serials have simpler frequency components are easier to be forecasted. So has...
Abstract The task of remote sensing image object detection in low latency scenes is great research significance. To address the problem that current high‐precision algorithm based on a feature pyramid network slow due to large number parameters and complicated computation, fast method Single‐scale Feature Transformation Network (SFTN) proposed. Firstly, single‐scale features, new channel features are quickly generated by linear transformation original convolution kernel clustering...
This article presents a sensor-less control strategy of Surface Permanent Magnet Synchronous Motor (SPMSM) in full speed operations. To realize closed-loop currents and velocity middle-to-high speed, Sliding Mode Observer (SMO) is designed to estimate the rotor position which provide for Field-Oriented Control (FOC). suppress high frequency chattering induced by signum function, an improved SMO based on saturation function law adopted. Because back-electromotive force extract failed start up...
In order to compensate for the disadvantages of large computational scale and high requirements chip arithmetic traditional extended Kalman filter, reduce volume improve stability motor controller, dependence algorithm on while ensuring control accuracy. An filter system based weighted average fusion is designed. A without position sensor also designed, which can complete with significantly reduced load main controller chip, has robustness anti-interference capability. Finally, effectiveness...
As the application of large scale information network, mount content in network is increased sharply. This brings challenge to technology searching and monitoring content. In this article, it introduced an locating strategy, which based on multi-agent can be distributed work parallel. The concept search-degree search-rewards are for agents overall system simultaneously. object search numerous nodes or Websites at a very fast speed not needing occupying too much resources. At same time, some...