A5042878427- Advanced Memory and Neural Computing
- Ferroelectric and Negative Capacitance Devices
- Parallel Computing and Optimization Techniques
- Neural dynamics and brain function
- Neural Networks and Reservoir Computing
- Embedded Systems Design Techniques
- Distributed and Parallel Computing Systems
- Neuroscience and Neural Engineering
- Neural Networks and Applications
- Embedded Systems and FPGA Design
- Interconnection Networks and Systems
- EEG and Brain-Computer Interfaces
- Cloud Computing and Resource Management
- Cloud Computing and Remote Desktop Technologies
- Quantum Computing Algorithms and Architecture
- IoT-based Smart Home Systems
- Simulation Techniques and Applications
- Metaheuristic Optimization Algorithms Research
- IoT and Edge/Fog Computing
- Video Surveillance and Tracking Methods
- Computability, Logic, AI Algorithms
- Distributed Control Multi-Agent Systems
- Caching and Content Delivery
- Artificial Immune Systems Applications
- Stochastic Gradient Optimization Techniques
Tsinghua University
2013-2025
NARI Group (China)
2023
National Engineering Research Center for Information Technology in Agriculture
2020
Tianjin University of Technology
2016
Northwestern Polytechnical University
2009
With the recent reincarnations of neuromorphic computing comes promise a new paradigm, with focus on design and fabrication chips. A key challenge in design, however, is that programming such chips difficult. This paper proposes systematic methodology set tools to address this challenge. The proposed toolset called NEUTRAMS (Neural network Transformation, Mapping Simulation), includes three components: neural (NN) transformation algorithm, configurable clock-driven simulator an optimized...
With the recent reincarnations of neuromorphic computing comes promise a new paradigm, with focus on design and fabrication chips. A key challenge in design, however, is that programming such chips difficult. This paper proposes systematic methodology set tools to address this challenge. The proposed toolset called NEUTRAMS (Neural network Transformation, Mapping Simulation), includes three components: neural (NN) transformation algorithm, configurable clock-driven simulator an optimized...
Sparse-Matrix Dense-Matrix Multiplication (SpMM) and Sampled Dense Matrix (SDDMM) are important sparse kernels in various computation domains. The uneven distribution of nonzeros the matrix tight data dependence between dense matrices make it a challenge to run multiplication efficiently on GPUs. By analyzing aforementioned problems, we propose row decomposition (RoDe)-based approach optimize two GPUs, using standard Compressed Sparse Row (CSR) format. Specifically, RoDe divides rows into...
Many believe the future of gaming lies in cloud, namely Cloud Gaming, which renders an interactive application cloud and streams scenes as a video sequence to player over Internet. This paper proposes GCloud , GPU/CPU hybrid cluster for based on user-level virtualization technology. Specially, we present performance model analyze server-capacity games’ resource-consumptions, categorizes games into two types: CPU-critical memory-io-critical . Consequently, several scheduling strategies have...
Spiking neural network (SNN) is the most commonly used computational model for neuroscience and neuromorphic computing communities. It provides more biological reality possesses potential to achieve high power energy efficiency. Because existing SNN simulation frameworks on general-purpose graphics processing units (GPGPUs) do not fully consider oriented properties of SNNs, like spike-driven, activity sparsity, etc., they suffer from insufficient parallelism exploration, irregular memory...
Brain-inspired computing refers to computational models, methods, and systems, that are mainly inspired by the processing mode or structure of brain. A recent study proposed concept "neuromorphic completeness" corresponding system hierarchy, which is helpful determine capability boundary brain-inspired judge whether hardware software compatible with each other. As a position paper, this article analyzes existing chips' design characteristics current so-called "general purpose" application...
Spiking Neural Networks (SNNs) are currently the most widely used computing model for neuroscience communities. There is also an increasing research interest in exploring potential of SNN brain-inspired computing, artificial intelligence, and other areas. As SNNs possess distinguished characteristics that originate from biological authenticity, they require dedicated simulation frameworks to achieve usability efficiency. However, there no widely-used, easily accessible, high performance...
The Ant Task Allocation algorithm is proposed for task allocation in multi-agent systems, which inspired by the swarm intelligence of social insects. a variation Colony Optimization, selection model honeybees adopted. In simulation experiments, achieves efficient and reasonable random working environment. Moreover, when condition changes, implements effective re-allocation. Experimental results indicate adaptability robustness algorithm.
This paper introduces the design and implementation measures of network video surveillance system based on embedded system. chooses S3C2440 which ARM9 architecture as an processor uses stable efficient Linux its operating Based ordinary remote system, a new algorithm for detecting moving abject is added, Intrusion Detection realized by kind adaptive background subtraction algorithm. When invasion occurs, GPRS will send alarm message. Users can get realtime data through way browsing web. In...
Abstract Computational neural models are essential tools for neuroscientists to study the functional roles of single neurons or circuits. With recent advances in experimental techniques, there is a growing demand build up at neuron large-scale circuit levels. A long-standing challenge such lies tuning free parameters closely reproduce recordings. There many advanced machine-learning-based methods developed recently parameter tuning, but them task-specific requires onerous manual...
In the architecture of large screen display system, graphic workstation is mainly used as carrier to project content on through splicing controller. When fails, it will directly affect and effect screen. This paper proposes a disaster recovery scheme for that most prone failure in system architecture. software method switch data transmission channel failed standby workstation, so ensure can be automatically shielded shortest time quickly return normal state.
Abstract Inferring the monosynaptic connectivity of neural circuits from in vivo experimental data is essential for understanding architecture that underpins behavior and cognition. However, advanced machine learning (ML) methods, especially deep learning, face significant challenges because observation limited incomplete, making it impractical to obtain ground-truth labeling. As a result, researchers typically rely on synthetic generated by biophysical models initial training. this reliance...