ReGraphX: NoC-enabled 3D Heterogeneous ReRAM Architecture for Training Graph Neural Networks.
Aqeeb Iqbal ArkaJanardhan Rao DoppaPartha Pratim PandeBiresh Kumar JoardarKrishnendu ChakrabartyPublished in: DATE (2021)
Keyphrases
- neural network
- training process
- multi processor
- training algorithm
- neural network structure
- network architecture
- feedforward neural networks
- multi layer
- feed forward neural networks
- neural network training
- network on chip
- genetic algorithm
- graph representation
- multi layer perceptron
- graph theory
- backpropagation algorithm
- back propagation
- pattern recognition
- multilayer perceptron
- error back propagation
- multilayer neural network
- management system
- learning capabilities
- neural network model
- directed graph
- heterogeneous environments
- artificial neural networks
- packet switched
- loosely coupled
- feedforward artificial neural networks
- weighted graph
- graph structure
- associative memory
- recurrent neural networks
- feed forward
- structured data
- random walk
- fuzzy logic
- activation function
- design methodology
- directed acyclic graph
- radial basis function network
- cmos technology
- high speed
- training set
- training data