Non-Centered Parameterisations for Hierarchical Models and Data Augmentation

By J. Bernardo, Maria J. Bayarri, James O. Berger, A. P. Dawid, David E. Heckerman, A. F. Smith, M. West. 2003

Subfields: Probabilistic & Bayesian Methods

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Probabilistic & Bayesian Methods

Citations in the corpus

Variational Inference with Normalizing Flows (2015, methodology): “…continuous latent variables, and the approach we take computes the required gradients us-ing a non-centered reparameterization of the expectation (Papaspiliopoulos et al., 2003; Williams, 1992), combined with Monte Carlo approximation — referred to as stochastic backpropagation (Rezende et al.,…”
Analysis of the Gibbs Sampler for Hierarchical Inverse Problems (2013, methodology): “We have considered the interweaving method of [41], which combines in each iteration centered and noncentered draws of δ, and the partially noncentered parametrizations of [33], in which the prior is reparametrized as u = δ− t 2 vt, where vt ∼ N (0, δt−1C0) for some t ∈ [0, 1].”
Pathfinder: Parallel quasi-Newton variational inference (2021, unknown): “The posteriordb package includes the eight schools model of Rubin (1981), in both a centered and non-centered parameterization; see Papaspiliopoulos et al. (2003); Stan Development Team (2021b); Betancourt and Girolami (2015) for more information on these parameterizations.”
To Center or Not to Center: That Is Not the Question—An Ancillarity–Sufficiency Interweaving Strategy (ASIS) for Boosting MCMC Efficiency (2011, methodology): “Section 2 defines the ASIS and more generally the component-wise ASIS, and uses examples from the works of Liu and Wu (1999) and Papaspiliopoulos, Roberts, and Skold (2003, 2007) to show how, why, and when ASIS works.”
Particle gibbs with ancestor sampling (2014, background): “This can result in a significant improvement in mixing of the Gibbs sampler, in particular when there are strong dependencies between the system state and the parameters (Golightly and Wilkinson, 2008; Papaspiliopoulos et al., 2003).”
New inference strategies for solving Markov Decision Processes using reversible jump MCMC (2009, methodology): “This is an idea closely related to the techniques discussed by (Papaspiliopoulos et al., 2003).”
Asymptotically exact inference in differentiable generative models (2016, background): “Whether it is necessarily helpful to remove dependencies in pu like this for the proposed method is an open question and will likely be model specific; it has previously been found that non-centred parameterisations can be beneficial when performing MCMC inference in hierarchical models when the uno”
Scaling limits for the transient phase of local Metropolis–Hastings algorithms (2005, methodology): “In fact, theory predicts (see Roberts and Rosenthal, 1998) that Langevin algorithms will almost always massively outperform Metropolis alternatives in the stationary phase, whenever they are practically feasible to run. For further examples of the use of Langevin methods for high-dimensional Bayesia”
scCODA is a Bayesian model for compositional single-cell data analysis (2021, methodology): “Effects between covariates m and cell types k are hierarchically modeled using individual, normally distributed effects γ m ; k with a covariate-speci ﬁ c scaling factor σ 2 m 29,30 .”
Bayesian inference for non‐Gaussian Ornstein–Uhlenbeck stochastic volatility processes (2004, methodology): “The methodology that we introduce is related to the notion of non-centring, as described in Papaspiliopoulos (2003) and Papaspiliopoulos et al. (2003), and can be applied in contexts outside this paper.”