Theoretical and Practical Advances in Hierarchical Bayesian Deep Learning Methodology
This work introduces the multilevel particle marginal Metropolis-Hastings (MLPMMH) method, which combines pseudo-marginal inference for the hyperparameters with coupled sequential Monte Carlo (SMC) across hierarchy of network-width truncations for conditional parameter inference.
N. K. Chada, S. Cotter, K. J. H. Law et al.
· 0 citations