Liberty or Depth: Deep Bayesian Neural Nets Do Not Need Complex Weight Posterior Approximations.
Sebastian FarquharLewis SmithYarin GalPublished in: NeurIPS (2020)
Keyphrases
- neural nets
- neural network
- posterior distribution
- posterior probability
- feed forward
- backpropagation neural networks
- learning tasks
- higher order
- depth map
- counter propagation
- maximum likelihood
- probabilistic model
- artificial neural networks
- single layer
- deep architectures
- decision trees
- kernel function
- real time
- bayesian inference
- bayesian networks
- belief nets