| Description |
-
Bayesian inference is a powerful framework to do prediction under
uncertainty. Bayesian Neural Networks combine the flexibility of neural networks with
Bayesian machine learning's ability to incorporate uncertainty into prediction. While
approximate Bayesian makes restricting assumption on the class of posterior
distribution, particle-based algorithms can learn more flexible posterior distributions.
... read more The following thesis outlines, compares and contrast two particle-based algorithms for
Bayesian Neural Network: Hamiltonian Monte Carlo and Stein Variational Gradient Descent.
It then proposes a third algorithm that is a hybrid of the two aforementioned
algorithms.read less
|
|---|
This object is in collection
