Sharpness-Aware Training for Accurate Inference on Noisy DNN Accelerators.
Gonçalo MordidoSarath ChandarFrançois Leduc-PrimeauPublished in: CoRR (2022)
Keyphrases
- training process
- structured prediction
- computationally efficient
- supervised learning
- training data
- highly accurate
- graphical models
- belief networks
- probabilistic inference
- training set
- inference engine
- training phase
- neural network
- training examples
- belief propagation
- noisy data
- general purpose
- bayesian inference
- training algorithm
- visual perception