A5040566204- Advanced Neural Network Applications
- CCD and CMOS Imaging Sensors
- Leech Biology and Applications
- Neural Networks and Applications
- Muscle activation and electromyography studies
- Hand Gesture Recognition Systems
- Ginkgo biloba and Cashew Applications
- Domain Adaptation and Few-Shot Learning
- Infrared Target Detection Methodologies
- Advanced Text Analysis Techniques
- User Authentication and Security Systems
- Neurological Disease Mechanisms and Treatments
- Antibiotic Resistance in Bacteria
- Coastal wetland ecosystem dynamics
- Brain Tumor Detection and Classification
- Intelligent Tutoring Systems and Adaptive Learning
- Fault Detection and Control Systems
- Food Quality and Safety Studies
- Bacterial biofilms and quorum sensing
- Environmental Toxicology and Ecotoxicology
- Salmonella and Campylobacter epidemiology
- Advanced Sensor and Energy Harvesting Materials
- Marine and coastal plant biology
- Advanced Memory and Neural Computing
- Information Retrieval and Search Behavior
South China University of Technology
2023-2024
Beijing Microelectronics Technology Institute
2022-2023
Beijing University of Chinese Medicine
2021-2022
Center for Disease Control
2021
Ocean University of China
2016
In real-time human-machine interaction (HMI) applications, hand gesture recognition (HGR) requires high accuracy with low latency. Surface electromyography (sEMG), a physiological electrical signal reflecting muscle activation, is extensively used in HMI. Recently, transient sEMG, generated during the transitions, has been employed HGR to achieve lower observational latency compared steady-state sEMG. However, use of long feature windows (up 200 ms) still make it less desirable low-latency...
The demand for adopting neural networks in resource-constrained embedded devices is continuously increasing. Quantization one of the most promising solutions to reduce computational cost and memory storage on devices. In order complexity overhead deploying Integer-only hardware, current quantization methods use a symmetric mapping strategy quantize floating-point network into an integer network. However, although has advantage easier implementation, it sub-optimal cases where range could be...
In recent years, high-density surface electromyography (HD-sEMG) has been applied for cancelable authentication of wearable devices used in healthcare applications. However, previous studies focused on the forearm or hand, whereas wrist is more comfortable practical use. Covering a larger area with fewer electrodes (16 channels versus 256 forearm) also reduces chance electrode corruption and improves stability system. addition, fast dynamic gestures are desirable than static real-time...
With the increasing complexity of space missions, demand for intelligent technologies in aerospace is rapidly growing. Neural network accelerators play an indispensable role applications as dedicated hardware devices accelerating neural inference. In this paper, we have explored challenges associated with selecting appropriate Processing Unit (NPU) applications. We shown that peak specifications NPU may not accurately represent its actual performance. To overcome challenge, conducted a...
The content of total flavonol glycosides in Ginkgo Folium the planting bases was determined by high performance liquid chromatography(HPLC).The samples were extracted reflux with methanol-25% hydrochloric acid.The HPLC conditions as follows: Agilent ZORBAX SB-C_(18) column(4.6 mm×250 mm, 5 μm), isocratic elution mobile phase 0.4% phosphoric acid solution-methanol(45∶55), flow rate 1 mL·min~(-1), column temperature 30 ℃, detection wavelength 360 nm, and injection vo-lume 10 μL.A method for...
The demand for adopting neural networks in resource-constrained embedded devices is continuously increasing. Quantization one of the most promising solutions to reduce computational cost and memory storage on devices. In order complexity deploying Integer-only hardware, current quantization methods use a symmetric mapping strategy. However, robustness generalization this strategy are poor. It difficult quantized model meet accuracy requirements, especially complicated tasks with higher...
This paper uses the inverted soft budget constraint theory to analyze relationship between private capital local government and financial innovation. The participation of Private Capital financing is result under economic crisis. As crisis mitigation, try harden by innovation which in deed reintroduction market mechanism. In order encourage innovation, this also focus on emphasizing importance developing third party institutions, build filter fusion function institutions government. By way...
Spiking Neural Networks (SNNs) is regarded as the next generation of Artificial (ANNs) and an alternative to Convolutional (CNNs). The CNN– SNN conversion considered one most promising approaches generate SNNs. However, previous works showed that methods required a long timesteps (also known inference latency in CNNs) get acceptable accuracy loss. On other hand, rarely used for object detection. In this paper we proposed Leaky-Integrate-and-Fire (LIF) neuron model which can complete...
The more intelligent and complex model has brought huge computing demand high Memory bandwidth. It is deployed on the embedded neural network accelerator meets need to reduce bit width in real time capacity requirements of devices. With expansion task scenarios, must through quantitative perception training, but it will also introduce expensive training costs, making Toolchain bloated complex, which not conducive rapid new scenarios. To address this problem, a task-agnostic knowledge...
The deployment of neural networks on the embedded devices is an important development trend in future. Due to limited resources devices, acceleration urgent and great importance. Quantization, which converts floating-point into low-bit-width integer networks, one most promising solutions for reducing computing storage cost devices. Concat a common operator networks. Since all branches that are being concatenated should share same quantization parameters, results large loss accuracy. It...
Whitmania pigra Whitman is a non-bloodsucking leech, which included in the Pharmacopoeia of People's Republic China. As traditional Chinese medicine, its main function to remove blood stasis. In process breeding, easily affected by pollution heavy metal lead. This study simulated pond culture environment aquarium with soil bottom. The lead contamination was caused adding nitrate soil. blank leech group and contaminated were obtained after 50 days culturing aquariums. Denovo assembly analysis...
Abstract Background: With the rapid emergence of plasmid mediated polymyxin resistance gene mcr-1 , multidrug-resistant Salmonella caused great troubles in clinical treatment and attracted extensive attention. Here we report strains harboring China, which are resistant to both polymyxin, traditional antibiotics even clinically wide-used antibiotics. Methods: We screened 1454 collected our laboratory from 2006 2018 3 provinces or regions for by PCR. Antimicrobial susceptibility testing was...
In this experiment, an ultra-high performance liquid chromatographytandem triple quadrupole mass spectrometry was established for the determination of caffeine in commercially available Ginkgo Folium. The samples were extracted by ultrasonic method with methanol, and separated on Waters CORTECS T3 column(2.1 mm×100 mm, 2.7 μm), mobile phase 0.1% formic acid solution-0.1% acetonitrile solution gradient elution, at flow rate 0.3 mL·min~(-1); column temperature 30 ℃, injection volume 2 μL. Mass...
Whitmania pigra is the most widely distributed species of leeches in market. In this study, effect heavy metal lead pollution on anticoagulant activity Wh. was studied and potential mechanism explored. Pb(NO_3)_2 used to contaminate breeding soil which then rear for 50 days(lead-contaminated group, LC group), meanwhile blank control group(CG group) set. Proteins were extracted from obtained leech samples, differentially expressed proteins between CG groups analyzed by label-free proteomics...