A5100452580- Advanced Graph Neural Networks
- Advanced Computational Techniques and Applications
- Cryptography and Data Security
- Advanced Neural Network Applications
- Topic Modeling
- Ferroelectric and Negative Capacitance Devices
- Advanced Sensor and Control Systems
- Embedded Systems Design Techniques
- Interconnection Networks and Systems
- Privacy-Preserving Technologies in Data
- Radiomics and Machine Learning in Medical Imaging
- Medical Image Segmentation Techniques
- Parallel Computing and Optimization Techniques
- Transportation and Mobility Innovations
- Advanced Memory and Neural Computing
- Iterative Methods for Nonlinear Equations
- VLSI and Analog Circuit Testing
- Mobile Crowdsensing and Crowdsourcing
- Data Quality and Management
- Neural Networks and Applications
- Privacy, Security, and Data Protection
- Network Security and Intrusion Detection
- Blockchain Technology Applications and Security
- Reinforcement Learning in Robotics
- Smart Grid and Power Systems
Qingdao University of Technology
2025
South China Institute of Collaborative Innovation
2023
Chinese Academy of Sciences
2020-2022
University of Chinese Academy of Sciences
2018-2022
Institute of Computing Technology
2020-2022
North China University of Technology
2022
Beijing Institute of Big Data Research
2022
Suzhou University of Science and Technology
2022
University of Science and Technology of China
2022
Hebei Normal University of Science and Technology
2022
Graph analytics is an emerging application which extracts insights by processing large volumes of highly connected data, namely graphs. The parallel graphs has been exploited at the algorithm level, in turn incurs three irregularities onto computing and memory patterns that significantly hinder efficient architecture design. Certain can be partially tackled prior domain-specific accelerator designs with well-designed scheduling data access, while others remain unsolved.
The thermal runaway propagation (TRP) model of energy storage batteries can provide solutions for the safety protection systems. Traditional TRP models are solved using finite element method, which significantly consume computational resources and time due to large number elements nodes involved. To ensure solution accuracy improve efficiency, this paper transforms heat transfer problem in calculations into a state-space equation form based on reduced-order theory linear time-invariant (LTI)...
In recent years, with the rapid development of Internet technology and applications, scale data has exploded, which contains a significant amount valuable knowledge. The best methods for organization, expression, calculation, deep analysis this knowledge have attracted great deal attention. graph emerged as rich intuitive way to express Knowledge reasoning based on graphs is one current research hot spots in played an important role wireless communication networks, intelligent question...
With the convergence of fixed and mobile networks, heterogeneous networks are becoming ubiquitous. Internet giants seeing plight identity authentication. To address this issue, unified access management (UAM) was conceived. This paper provides a novel scheme, named SGX-UAM, with one-time passwords (OTPs) based on Intel software guard extensions (SGX). SGX-UAM outperforms generic UAM for providing resistance to most client attacks, man-in-the-middle (MITM) phishing replay attacks denial...
With the development of cloud computing, software system architecture has gradually changed from a single to service-oriented architecture, which microservice is typical representative. It committed providing users with more reliable, maintainable, and extensible design services. Although many advantages, because there are multiple services in it becomes difficult detect faults when fails. How efficiently causes key technology ensure performance reliability microservices. Aiming at this...
Preserving the confidentiality of sensitive data, while permitting knowledge discovery, is an important goal in privacy-preserving data mining. This paper investigates effectiveness shuffling for classification tree and regression analysis. We compare to based perturbation method which was developed specifically purpose Results suggest that provides higher levels security more effectively preserves mining than method.
In this paper, we consider task offloading in vehicular edge computing systems. The limited battery limits the mileage of electric vehicles. For a longer driving distance, vehicle can not only be performed locally, but also offload to save energy. way, energy consumption system is reduced and utilization rate whole improved. involves optimization variables which are where tasks frequency, coupled. So it hard decide how allocate tasks. We propose that solve problem by using genetic algorithm....
Limited by the memory capacity and compute power, singe-node graph convolutional neural network (GCN) accelerators cannot complete execution of GCNs within a reasonable amount time, due to explosive size graphs nowadays. Thus, large-scale call for multi-node acceleration system (MultiAccSys) like TPU-Pod networks. In this work, we aim scale up single-node GCN accelerate on graphs. We first identify communication pattern challenges observe that (1) coarse-grained patterns exist in...
Derived from the fusion of graph traversal and neural networks, convolutional networks (GCNs) have achieved state-of-the-art performance in learning. However, hybrid execution pattern, caused by opposite characteristics based phase network transformation phase, poses huge challenges to efficient traditional architectures. Although GCN accelerators emerged address these challenges, they fail harvest both bidirectional inter-phase opportunities exposed alternate phases GCNs. Previous works...
Dataflow processor has shown its unique advantages in executing high performance computing applications with communication-exposed microarchitecture. In dataflow processors, large amounts of data are directly transferred between instructions through a network-on-chip. The efficiency transfer is an imperative metric that needs to be optimized most processors. Based on the specific features network, we propose mechanism for dynamically merging packets routers. By testing workloads varying...
As a hot research direction in current academic studies, knowledge graph reasoning is aimed at solving the many challenges and pain points of graphs. This paper centers around temporal data prediction presents multi-level framework that leverages causal Our seamlessly integrates graphs with to enhance accuracy. The composed two key components: construction gated neural network prediction. By representing facts relationships within domain using graphs, enhances capability model. proposed...
Most e-commerce product feeds provide blended results of advertised products and recommended to consumers. The underlying advertising recommendation platforms share similar if not exactly the same set candidate products. Consumers' behaviors on constitute part model's training data therefore can influence results. We refer this process as Leverage. Considering mechanism, we propose a novel perspective that advertisers strategically bid through platform optimize their organic traffic. By...
In heterogeneous networks, the random walks based on meta-path requires prior knowledge and lacks flexibility. And non-meta-path only considers number of node types, but does not consider influence schema topology between types in real networks. To solve above problems, this paper proposes a novel model HNE-RWTIC (Heterogeneous Network Embedding Based Random Walks Type & Inner Constraint). Firstly, to realize flexible walks, we design strategy, which is type selection strategy...
Limited by the memory capacity and compute power, singe-node graph convolutional neural network (GCN) accelerators cannot complete execution of GCNs within a reasonable amount time, due to explosive size graphs nowadays. Thus, large-scale call for multi-node acceleration system (MultiAccSys) like TPU-Pod networks. In this work, we aim scale up single-node GCN accelerate on graphs. We first identify communication pattern challenges observe that (1) coarse-grained patterns exist in...