Particle gibbs with ancestor sampling

By Fredrik Lindsten, Michael I. Jordan, Thomas Bo Schön. 2014

279 citations recorded by the source corpus.

Particle Markov chain Monte Carlo (PMCMC) is a systematic way of combining the two main tools used for Monte Carlo statistical inference: sequential Monte Carlo (SMC) and Markov chain Monte Carlo (MCMC). We present a new PMCMC algorithm that we refer to as particle Gibbs with ancestor sampling (PGAS). PGAS provides the data analyst with an off-the-shelf class of Markov kernels that can be used to simulate, for instance, the typically high-dimensional and highly autocorrelated state trajectory in a state-space model. The ancestor sampling procedure enables fast mixing of the PGAS kernel even when using seemingly few particles in the underlying SMC sampler. This is important as it can significantly reduce the computational burden that is typically associated with using SMC. PGAS is conceptually similar to the existing PG with backward simulation (PGBS) procedure. Instead of using separate forward and backward sweeps as in PGBS, however, we achieve the same effect in a single forward sweep. This makes PGAS well suited for addressing inference problems not only in state-space models, but also in models with more complex dependencies, such as non-Markovian, Bayesian nonparametric, and general probabilistic graphical models.

Subfields: Artificial Intelligence, Statistics and Probability

How researchers see this paper

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

In Artificial Intelligence

In Statistics and Probability

Citations in the corpus

Causal Discovery from Temporally Aggregated Time Series (2017, methodology): “, 2014), which combines the efficient conditional particle filter with ancestor sampling (CPF-AS) (Lindsten et al., 2014) with the stochastic approximation EM (SAEM) algorithm (Delyon et al.”
Nested Sequential Monte Carlo Methods (2015, methodology): “lts, since the PMCMC distribution is deﬁned w.r.t. to n, whereas the backward simulator of Algorithm 6 works with the original target distribution ˇ n. Nevertheless, from the proof of Lemma 1 by Lindsten et al. (2014) it follows that we can write the following collapsed conditional distribution of”
Bayesian computation: a summary of the current state, and samples backwards and forwards (2015, methodology): “This approach is being used increasingly in complex dynamic models like those found in signal processing (Whiteley et al. 2010), dynamical systems like the PDEs in biochemical kinetics (Wilkinson 2011b) and probabilistic graphical models (Lindsten et al. 2014).”
Applications of Probabilistic Programming (Master's thesis, 2015) (2016, methodology): “In 2014 a general-purpose implementation of the particle Gibbs with ancestor sampling method (Lindsten et al., 2014) was introduced in (van de Meent et al.”
Bayesian computation: a perspective on the current state, and sampling backwards and forwards (2015, methodology): “This approach is being used increasingly in complex dynamic models like those found in signal processing (Whiteley et al. 2010), dynamical systems like the PDEs in biochemical kinetics (Wilkinson 2011b) and probabilistic graphical models (Lindsten et al. 2014).”
Interacting Particle Markov Chain Monte Carlo (2016, methodology): “Another option for the CSMC nodes is to replace this step by internal ancestor sampling (Lindsten et al., 2014) steps and simulate from the final weights as normal.”
Identification of jump Markov linear models using particle filters (2014, methodology): “More specifically we will make use of particle MCMC (PMCMC), which is a systematic way of exploring the strengths of both approaches by using SMC to construct the necessary high-dimensional Markov kernels needed in MCMC [4], [5].”