IMA-GNN: In-Memory Acceleration of Centralized and Decentralized Graph Neural Networks at the Edge.
Mehrdad MorsaliMahmoud NazzalAbdallah KhreishahShaahin AngiziPublished in: ACM Great Lakes Symposium on VLSI (2023)
Keyphrases
- neural network
- weighted graph
- peer to peer
- undirected graph
- auto associative
- peer to peer systems
- edge weights
- graph representation
- fuzzy logic
- vertex set
- pattern recognition
- graph model
- graph structure
- random walk
- disjoint paths
- neural network model
- bipartite graph
- directed graph
- connected components
- graph theory
- graph partitioning
- cooperative
- feed forward
- genetic algorithm
- back propagation
- artificial neural networks
- edge detection
- distributed environment
- strongly connected
- fully distributed
- multiscale
- limited memory
- memory usage
- mobile robot
- edge information
- graph theoretic
- distributed systems
- fuzzy systems
- multi agent
- memory size
- structured data
- fault diagnosis
- self organizing maps
- image processing
- associative memory