A5100447533- Advanced Memory and Neural Computing
- Neural dynamics and brain function
- 3D IC and TSV technologies
- VLSI and Analog Circuit Testing
- Neural Networks and Applications
- Integrated Circuits and Semiconductor Failure Analysis
- High Entropy Alloys Studies
- Ferroelectric and Negative Capacitance Devices
- stochastic dynamics and bifurcation
- Neural Networks and Reservoir Computing
- Neuroscience and Neural Engineering
- Opportunistic and Delay-Tolerant Networks
- Underwater Vehicles and Communication Systems
- Gait Recognition and Analysis
- Medicinal Plant Pharmacodynamics Research
- Video Surveillance and Tracking Methods
- VLSI and FPGA Design Techniques
- Aerogels and thermal insulation
- Microgrid Control and Optimization
- Hand Gesture Recognition Systems
- Silicone and Siloxane Chemistry
- Cancer-related molecular mechanisms research
- Smart Grid Energy Management
- Additive Manufacturing Materials and Processes
- Advanced Cellulose Research Studies
Xi'an Jiaotong University
2019-2023
Nanjing University of Information Science and Technology
2022
Tianjin University of Technology
2021
Lanzhou University of Technology
2019
The First Affiliated Hospital, Sun Yat-sen University
2018
Sun Yat-sen University
2018
Nanjing University of Aeronautics and Astronautics
2017
Duke University
2016
Nvidia (United States)
2016
ABSTRACT A novel and environmentally friendly attapulgite‐based aerogel with a three‐dimensional fibrillary network structure was prepared by incorporation of nanometer‐sized natural clay crystals, in this case attapulgite (ATP), into degradable poly(vinyl alcohol) (PVA), cotton cellulose nanowhisker, melamine using only simple blending method subsequently freeze‐drying process. The ATP‐based exhibits an abundant porosity average mesopore size 8.0 nm diameter. Compared to fragile rigid...
In this brief, a cost-efficient and scalable spiking convolution neural network architecture is proposed. Reusable modules are designed to reduce hardware resource consumption by leveraging the structural similarity of each convolutional layer pooling in (SCNN). Taking advantage time-driven processing, memory reutilization applied for storing states. Benefit from these reusable modules, proposed demonstrates outstanding scalability. Two SCNN structures (named I II) with different...
Sparse convolutional neural network (CNN) accelerators face challenges such as low utilization of processing elements (PEs), data reuse, and high hardware sparse index addressing cost during convolution operations. This brief proposes a new scheduling strategy for feature map vectors kernel based on the systolic array row stationary (RS) dataflow to overcome these problems. Our approach employs design implementation operation module array, simulated comprehensively in Vivado using Virtex-7...
In this brief, an event-driven spiking convolution architecture with multi-kernel and multi-layer capability is designed. The proposed can be configured in multiple-spike (MS) mode or single-spike (SS) to adapt different neural network (SCNN) models either rate coding scheme temporal scheme. A skipped zero kernel step designed reduce access neuron membrane potentials memories. addition, the design supports two pooling methods for increasing flexibility. implemented a Xilinx Kintex-7 FPGA...
Three-dimensional (3D) integration using through-silicon vias (TSVs) promises higher levels in a single package, keeping pace with Moore's law. Despite the promise and benefits offered by 3D integration, testing remains major obstacle that hinders its widespread adoption. This paper examines hype, myths, realities of IC testing. We describe number DfT challenges, present some solutions being advocated for challenges “What to Test”, “How “When Test”....
Interposer-based 2.5D integrated circuits (ICs) are seen today as a precursor to 3D ICs based on through-silicon vias (TSVs). All the dies and interposer in IC must be adequately tested for product qualification. This work provides solutions new challenges related testing of ICs. We propose test architecture using e-fuses pre-bond testing. design that is fully compatible with IEEE 1149.1 standard relies an enhancement access port (TAP) controller. present efficient built-in self-test (BIST)...
Interposer-based 2.5D integrated circuits (ICs) are seen today as a precursor to 3D ICs based on through-silicon vias. This paper describes some of the major challenges related testing and presents solutions these problems. We first describe test architecture using e-fuses for pre-bond interposer testing. next present an efficient built-in self-test (BIST) technique that targets dies interconnects. Finally, we programmable method shift-clock stagger assignment reduce power supply noise...
The spherical multi-robot system is mainly used for water quality monitoring in large-scale aquaculture. main research content of this article how to realize the effectiveness data transmission. Or, when detection range too large, adopts centralized control mode, once instability central robot network occurs. If can not be fed back user, ensure ability entire system. Novel nod has advantages Ad Hoc network, self - repair and limited by distance router. It been widely various fields. At same...
Spiking neural networks (SNNs) are artificial network models that closely mimic natural networks. LIF (Leaky Integrate-and-fire) neuron model, population coding and Tempotron supervised learning rules used to construct a spiking for visual color feature classification based on RGB-HSV (Red, Green, Blue -Hue, Saturation, Value) model. The product of momentum rate the last weight change is proposed speed up training SNN. Test results common data set show accuracy SNN can be 90%.
A method of human identification using micro-doppler data based on a lightweight neural network is proposed. It can identity various gaits and different person effectively, the amount computation number parameters model be reduced also. The principle like that used in linear phase finite impulse response filter digital signal processing utilized here. proposed LPCP-Conv kernel lower storage cost computational complexity model. results experiments show algorithm decrease quality as well...