Depth induces scale-averaging in overparameterized linear Bayesian neural networks.
Jacob A. Zavatone-VethCengiz PehlevanPublished in: CoRR (2021)
Keyphrases
- neural network
- morphological scale space
- pattern recognition
- bayesian networks
- fuzzy logic
- genetic algorithm
- highly non linear
- small scale
- recurrent neural networks
- feed forward
- neural network model
- depth map
- closed form
- posterior probability
- artificial neural networks
- scale invariant
- scale spaces
- three dimensional
- back propagation
- model selection
- scale space
- expert systems
- multi layer
- linear systems
- high quality