GraNDe: Near-Data Processing Architecture With Adaptive Matrix Mapping for Graph Convolutional Networks.
Sungmin YunByeongho KimJaehyun ParkHwayong NamJung Ho AhnEojin LeePublished in: IEEE Comput. Archit. Lett. (2022)
Keyphrases
- data processing
- network architecture
- graph theory
- adjacency matrix
- average degree
- management system
- graph structures
- data analysis
- fully connected
- data acquisition
- learning capabilities
- graph representation
- connected components
- dynamic networks
- community discovery
- graph structure
- real time
- structured data
- random walk
- data management
- overlapping communities
- low rank
- transition matrix
- network analysis
- singular value decomposition
- graph partitioning
- covariance matrix
- edge weights
- small world
- eigenvalues and eigenvectors
- laplacian matrix
- matrix representation
- graph layout
- graph model
- weighted graph
- graph matching
- pairwise
- bayesian networks
- social networks
- machine learning