GNN-PIM: A Processing-in-Memory Architecture for Graph Neural Networks.
Zhao WangYijin GuanGuangyu SunDimin NiuYuhao WangHongzhong ZhengYinhe HanPublished in: ACA (2020)
Keyphrases
- neural network
- processing elements
- associative memory
- memory management
- real time
- network architecture
- distributed processing
- multi layer
- artificial neural networks
- auto associative
- random access
- pattern recognition
- data processing
- software architecture
- parallel architecture
- computational power
- neural network model
- management system
- information processing
- graph model
- random walk
- weighted graph
- graph structure
- graph theory
- fuzzy logic
- back propagation
- main memory
- memory hierarchy
- platform independent
- connected components
- directed graph
- memory requirements
- multilayer perceptron
- recurrent neural networks
- hardware implementation
- graph representation
- graph theoretic
- data structure
- graph search
- graph matching
- database systems
- graph mining
- memory subsystem