Towards Training Billion Parameter Graph Neural Networks for Atomic Simulations.
Anuroop SriramAbhishek DasBrandon M. WoodSiddharth GoyalC. Lawrence ZitnickPublished in: ICLR (2022)
Keyphrases
- neural network
- training process
- training algorithm
- feed forward neural networks
- feedforward neural networks
- multi layer perceptron
- neural network training
- back propagation
- pattern recognition
- backpropagation algorithm
- error back propagation
- weighted graph
- multi layer
- parameter values
- graph theory
- artificial neural networks
- graph theoretic
- directed acyclic graph
- graph structure
- neural network model
- directed graph
- multilayer perceptron
- training patterns
- neural network structure
- graph partitioning
- multilayer neural network
- recurrent networks
- graph representation
- training phase
- graph model
- numerical simulations
- structured data
- training set
- spanning tree
- activation function
- fuzzy neural network
- feed forward
- test set
- training examples
- graph based algorithm
- random walk
- training data
- clustering algorithm