A domain-specific supercomputer for training deep neural networks.
Norman P. JouppiDoe Hyun YoonGeorge KurianSheng LiNishant PatilJames LaudonCliff YoungDavid A. PattersonPublished in: Commun. ACM (2020)
Keyphrases
- domain specific
- neural network
- training process
- training algorithm
- multi layer perceptron
- feed forward neural networks
- general purpose
- pattern recognition
- building blocks
- feedforward neural networks
- training examples
- error back propagation
- deep architectures
- neural network training
- domain independent
- domain experts
- back propagation
- artificial neural networks
- neural network structure
- backpropagation algorithm
- recurrent networks
- training phase
- training patterns
- training set
- information retrieval
- genetic algorithm
- massively parallel
- computer software
- data sets
- deep learning
- radial basis function network
- domain knowledge
- information extraction
- supervised learning
- neural network model
- online learning
- unsupervised learning
- test set