Shaving Weights with Occam's Razor: Bayesian Sparsification for Neural Networks Using the Marginal Likelihood.
Rayen DhahriAlexander ImmerBertrand CharpentierStephan GünnemannVincent FortuinPublished in: CoRR (2024)
Keyphrases
- marginal likelihood
- neural network
- model selection
- gaussian process
- closed form
- information criterion
- hyperparameters
- exponential family
- approximate inference
- expectation propagation
- back propagation
- cross validation
- bayesian learning
- probabilistic inference
- bayesian inference
- gaussian processes
- artificial neural networks
- parameter estimation
- probability model
- bayesian information criterion