ReGraphX: NoC-enabled 3D Heterogeneous ReRAM Architecture for Training Graph Neural Networks.
Aqeeb Iqbal ArkaBiresh Kumar JoardarJanardhan Rao DoppaPartha Pratim PandeKrishnendu ChakrabartyPublished in: CoRR (2021)
Keyphrases
- neural network
- training process
- multi processor
- training algorithm
- network architecture
- neural network structure
- network on chip
- error back propagation
- multi layer
- feed forward neural networks
- artificial neural networks
- back propagation
- pattern recognition
- loosely coupled
- feedforward neural networks
- backpropagation algorithm
- real time
- learning capabilities
- multilayer neural network
- directed graph
- management system
- heterogeneous systems
- recurrent networks
- graph representation
- training data
- graph model
- genetic algorithm
- graph theory
- graph structure
- bipartite graph
- training set
- recurrent neural networks
- feed forward
- routing algorithm
- packet switched
- fuzzy logic
- neural network model
- multilayer perceptron
- design methodology
- single processor
- heterogeneous environments
- weighted graph
- multi layer perceptron