A5089405414- Advanced Wireless Communication Techniques
- Advanced Memory and Neural Computing
- Error Correcting Code Techniques
- Advanced MIMO Systems Optimization
- CCD and CMOS Imaging Sensors
- Cellular Automata and Applications
- Target Tracking and Data Fusion in Sensor Networks
- Spectroscopy and Chemometric Analyses
- Antenna Design and Analysis
- Horticultural and Viticultural Research
- Antenna Design and Optimization
- Neural Networks and Applications
- Ferroelectric and Negative Capacitance Devices
- Microwave Engineering and Waveguides
- Energy Harvesting in Wireless Networks
- Neural Networks and Reservoir Computing
- Acute Ischemic Stroke Management
- Particle Detector Development and Performance
- Helicobacter pylori-related gastroenterology studies
- Medical Imaging Techniques and Applications
- Smart Agriculture and AI
- Bayesian Modeling and Causal Inference
- Visual Attention and Saliency Detection
- VLSI and FPGA Design Techniques
- Neurological Disease Mechanisms and Treatments
China Academy of Space Technology
2024
Southeast University
2020-2024
Tsinghua University
2023-2024
Purple Mountain Laboratories
2019-2023
Northeastern University
2018
The First People’s Hospital of Lianyungang
2014
Chinese Academy of Medical Sciences & Peking Union Medical College
2014
Health Commission of Anhui Province
2014
Ganzhou People's Hospital
2014
Liuyang City Maternal and Child Health Hospital
2014
Nowadays, artificial intelligence (AI) technology with large models plays an increasingly important role in both academia and industry. It also brings a rapidly increasing demand for the computing power of hardware. As AI continues to grow, growth hardware has failed keep up. This become significant factor restricting development AI. The augmentation is mainly propelled by escalation transistor density chip area. However, former impeded termination Moore's Law Dennard scaling, latter...
Global navigation satellite system (GNSS) array antenna receivers are widely used to suppress wideband interference in countermeasures. However, existing all adopt digital structure and beamforming technique, it has limited analog-front-end (AFE) dynamic range. In strong scenarios, AFE saturation will occur, which limits the maximum suppression ability of receiver. Aiming at this issue, paper proposes a robust method for GNSS based on hybrid technique. Firstly, novel fully connected receiver...
Massive multiple-input multiple-output (M-MIMO) brings better robustness and spectral efficiency but higher computational challenges compared to small-scale MIMO. One of the key is large-scale matrix inversion, as widely employed in channel estimation detection. Traditionally, address issue, several low-complexity inversion methods have been proposed, including tri-diagonal approximation (TMA) Neumann-series (NSA). Although previous effectively alleviate cost, they all fail exploit typical...
Expectation propagation (EP) achieves near-optimal performance for large-scale multiple-input multiple-output (L-MIMO) detection, however, at the expense of unaffordable matrix inversions. To tackle issue, several low-complexity EP detectors have been proposed. However, they all fail to exploit properties channel matrices, thus resulting in unsatisfactory non-ideal scenarios. this end, paper, a block-diagonal Neumann-series-based expectation approximation (BD-NS-EPA) algorithm is proposed,...
Cost-volume construction, which accurately computes the similarities between pixels in paired images, is a fundamental kernel of stereo vision processing and has been directly used robotic, autopilot, AR/VR applications. However, large parameter size consecutive data accesses real-time cost-volume construction (>30fps) exerts high demand on memory bandwidth (0.254Tb/s) operation (391GOPs). A promising candidate to resolve bottleneck computation-in-memory (CIM), provides computing parallelism...
Deep convolution neural networks (CNNs) usually require a large number of iterative operations, which would consume significant amounts hardware resources. In this paper, we propose an efficient architecture based on Winograd algorithm for convolutional (CNNs), by employing stochastic computing (SC). For the first step, fast algorithm, (WFCA), can lower complexity reducing multiplications is proposed. Although (SC) achieve reduction in compared with deterministic design, its straightforward...
By utilizing stochastic computing (SC), the hardware consumption of convolutional neural networks (CNNs) can be decreased significantly. However, long stream length is required to produce acceptable results, which leads extended computation time. As a result, inherent random fluctuation error and latency processing bitstreams have made previous SC-CNN implementations inefficient compared with conventional binary designs. To address these issues, in this brief, an efficient convolution...
Stroke system of care plays key roles both in providing effective therapies and improving the overall outcome patients with stroke. Our purpose was to develop evaluate Chinese rural areas.A stroke developed from November 2009 2010 3 townships Ganyu County. An additional matched were invited as controls. We first investigated management these then implemented an education campaign intervention townships. The effectiveness evaluated.There 1036 new among 344 345 subjects 6 communities....
The wireless baseband processing algorithms are still developing and show a great diversity. development of ASIC implementations cannot quickly adapt to the evolution standards. Meanwhile, general-purpose processors meet real-time requirements in some scenarios. This paper proposes DSP-purposed REconfigurable Acceleration Machine (DREAM) core for digital signal processing, which has good trade-off between flexibility performance. First, we abstract set shared operators with moderate...
This letter proposes an iterative multiple-input multiple-output (MIMO) receiver for polar-coded system using minimum mean-square error parallel interference cancellation (MMSE-PIC) algorithm. And we propose a new kind of tree-structured Gray codes to reduce the computational complexity MMSE-PIC without performance degradation. Simulation results indicate that, 64 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math...
A switched mode power supply (SMPS) with full digital control has many advantages over analog or mixed-signal implementation. This paper presents a controlled SMPS consisting of an AC-DC factor correction converter and phase-shifted full-bridge DC-DC converter. Digital algorithm PFC control, system design PSFB (phase shifted bridge) part, PWM strategy for are discussed. 500 W prototype been built tested under different conditions
Global navigation satellite system (GNSS) array antenna receivers are widely used to suppress wideband interference in countermeasures. However, existing all adopt a digital structure and beamforming technique, they have limited analog-front-end (AFE) dynamic range. In strong scenarios, AFE saturation will occur, which limits the maximum suppression ability of receiver. Aiming at this issue, paper proposes robust method for GNSS based on hybrid technique. Firstly, novel, fully connected...
Deep learning (DL) models are piquing high interest and scaling at an unprecedented rate. To this end, a handful of tiled accelerators have been proposed to support such large-scale training tasks. However, these often incorporate numerous cores or tiles even extending wafer-scale, substantial on-chip bandwidth, distributed memory systems. This results in exceedingly complex design space. Moreover, conducting actual experiments find optimal configurations is impractical due time constraints....
Benefiting from the self-attention mechanism, Transformer models have attained impressive contextual comprehension capabilities for lengthy texts. The requirements of high-throughput inference arise as large language (LLMs) become increasingly prevalent, which calls large-scale token parallel processing (LTPP). However, existing dynamic sparse accelerators struggle to effectively handle LTPP, they solely focus on separate stage optimization, and with most efforts confined computational...