A5075932308- Advanced Memory and Neural Computing
- Advanced Combustion Engine Technologies
- Biodiesel Production and Applications
- Advanced Neural Network Applications
- Lubricants and Their Additives
- Semiconductor materials and devices
- Ferroelectric and Negative Capacitance Devices
- Photonic and Optical Devices
- Image Processing Techniques and Applications
- COVID-19 and Mental Health
- Advanced Technologies in Various Fields
- Heat transfer and supercritical fluids
- Machine Learning and ELM
- Winter Sports Injuries and Performance
- Advanced Fluorescence Microscopy Techniques
- CCD and CMOS Imaging Sensors
- Heart Rate Variability and Autonomic Control
- Embedded Systems and FPGA Design
- Advanced Computing and Algorithms
- Shoulder Injury and Treatment
- Online Learning and Analytics
- Parallel Computing and Optimization Techniques
- Embedded Systems Design Techniques
- Advanced SAR Imaging Techniques
National University of Defense Technology
2022-2024
Central China Normal University
2022-2024
Kunming University of Science and Technology
2022-2023
Wuhan University of Technology
2020
The RRAM-based accelerators have become very popular candidates for neural network acceleration due to they perform matrix-vector multiplication in-memory with high storage density and low latency. Many related works used fixed-precision quantization achieve the model compression enhance tolerance of process variation, but these methods still suffer a large accuracy degradation poor robustness nonideal effects. In this work, we propose crossbar-aware mixed-precision scheme, which enables...
Background: Online learning is currently adopted by educational institutions worldwide to provide students with ongoing education during the COVID-19 pandemic. However, online has seen lose interest and become anxious, which affects performance leads dropout. Thus, measuring students’ engagement in imperative. It challenging recognize due lack of effective recognition methods publicly accessible datasets. Methods: This study gathered a large number videos at normal university. Engagement...
The residual carbon value and existent gum content were used to evaluate the deposition tendency of jatropha biodiesel with different oxidation degrees. chemical components analyzed before after by Gas chromatography-mass spectrometry (GC-MS) infrared spectrometer. results showed that in short-term process, both decreased first, then increased time. In long-term these two indicators monotonically from 1140 mg/100 mL 11,300 0.0272% 0.277%, respectively. After oxidation, aromatic compounds as...
Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI
Convolutional neural networks have different channel pruning sensitivities for layers, and the existing network methods cannot perform adaptive of channels within each layer according to sensitivity, resulting in irreversible damage during process. To solve this problem, a method with adjustment rate is proposed paper, which disassembles dynamic process into an iterative optimization problem finally obtains structure high compression accuracy. This performs under supervision loss constraint...
Abstract Combined with physical modelling, motion capture and microprocessor technology, a human knee joint intelligent monitoring system based on nine-axis attitude angle sensor WT901 is developed. The overall design of the its hardware software are completed. adopts STM32 to improve practicability system, use App enhance human-computer interaction capability. results tests indicate that realized accurate measurement knee-bending pressure. In addition, can also warn sports posture errors...
Binarized neural network (BNN) is a whose input activations and weights are quantified to 1-bit. It has very low parameters compared with other networks. What's more, the multiplication accumulation calculation of convolutional layer fully connected BNN can be replaced by XNOR-POPCOUNT calculation. FPGA become major hardware deployment platform for due its programmability parallelism. However, basic logic cell cannot directly complete This paper proposes an architecture that saves numerous...
With the continuous development of deep learning, Deep Convolutional Neural Network (DCNN) has attracted wide attention in industry due to its high accuracy image classification. Compared with other DCNN hardware deployment platforms, Field Programmable Gate Array (FPGA) advantages being programmable, low power consumption, parallelism, and cost. However, enormous amount calculation limited logic capacity FPGA restrict energy efficiency accelerator. The traditional sequential sliding window...
To explore the thermal conductivity of diesel-biodiesel-ethanol blended fuel under different temperatures and mixing ratios, three ternary fuels, namely, Jatropha/soybean/catering waste oil biodiesel-diesel-ethanol, were prepared. The biodiesel components analyzed using gas chromatography-mass spectrometry. A Hot Disk 2500S constant analyzer was used to measure fuels in temperature range 20–60°C. Thermal conductivities increased with increase temperature, had at same ratio. This may be...
The effects of light, ventilation and antioxidant on biodiesel performance the corresponding influence mechanisms are investigated during long-term storage. exhibits slow autoxidation its physical/chemical properties change Light conditions can accelerate oxidation biodiesel, thereby aggravating degradation. During process grinding spot diameter firstly decreases from 165.01 μm to 131.45 then increases 362.5 μm, similar shape a fishhook. initial storage stage, fatty acid methyl ester in is...