Wide Bayesian neural networks have a simple weight posterior: theory and accelerated sampling.
Jiri HronRoman NovakJeffrey PenningtonJascha Sohl-DicksteinPublished in: ICML (2022)
Keyphrases
- neural network
- posterior probability
- markov chain monte carlo
- posterior distribution
- theoretical framework
- monte carlo
- genetic algorithm
- wide range
- artificial neural networks
- bayesian networks
- general theory
- proposal distribution
- sequential monte carlo
- fuzzy systems
- image reconstruction
- back propagation
- parameter estimation
- particle filter
- probability distribution
- training data
- decision trees