Label Switching in Bayesian Mixture Models: Deterministic Relabeling Strategies

Rodríguez, Carlos E.; Walker, Stephen G.

doi:10.1080/10618600.2012.735624

Cited by 77 publications

(73 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The pivot is selected by choosing a high-posterior density point, such as the complete or non-complete Maximum A Posteriori (MAP) estimate. Rodriguez and Walker (2014) tried to relax the dependence of ECR algorithm to a pivot and proposed two iterative versions (ecr.iterative.1 and ecr.iterative.2). The first algorithm is using as input only the simulated allocation variables and is initialized by a pivot selected at random.…”

Section: Ecr Algorithmsmentioning

confidence: 99%

“…The data-based method of (Rodriguez and Walker 2014) is a deterministic relabeling algorithm. At first, a set of cluster centers m kr and dispersion parameters s kr is estimated for k = 1, .…”

Section: Data-based Relabelingmentioning

confidence: 99%

“…org/package=label.switching, which can be used in order to deal with the label switching problem using various algorithms that have been proposed to the related literature. More specifically, the label.switching package consists of eight relabeling methods: ordering constraints, the Kullback-Leibler based algorithm (Stephens 2000b), the pivotal reordering algorithm (Marin, Mengersen, and Robert 2005;Marin and Robert 2007), the default and iterative versions of ECR algorithm (Papastamoulis and Iliopoulos 2010;Rodriguez and Walker 2014;Papastamoulis and Iliopoulos 2013;Papastamoulis 2014), the probabilistic relabeling algorithm (Sperrin, Jaki, and Wit 2010) and the data-based algorithm (Rodriguez and Walker 2014).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

label.switching: AnRPackage for Dealing with the Label Switching Problem in MCMC Outputs

Papastamoulis¹

2016

J. Stat. Soft.

View full text Add to dashboard Cite

Label switching is a well-known and fundamental problem in Bayesian estimation of mixture or hidden Markov models. In case that the prior distribution of the model parameters is the same for all states, then both the likelihood and posterior distribution are invariant to permutations of the parameters. This property makes Markov chain Monte Carlo (MCMC) samples simulated from the posterior distribution non-identifiable. In this paper, the label.switching package is introduced. It contains one probabilistic and seven deterministic relabeling algorithms in order to post-process a given MCMC sample, provided by the user. Each method returns a set of permutations that can be used to reorder the MCMC output. Then, any parametric function of interest can be inferred using the reordered MCMC sample. A set of user-defined permutations is also accepted, allowing the researcher to benchmark new relabeling methods against the available ones.

show abstract

Section: Ecr Algorithmsmentioning

confidence: 99%

Section: Data-based Relabelingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

label.switching: AnRPackage for Dealing with the Label Switching Problem in MCMC Outputs

Papastamoulis¹

2016

J. Stat. Soft.

View full text Add to dashboard Cite

show abstract

“…Many methods exist on how to tackle the label switching problem, for example, there are methods that impose identifiability constraints (Diebolt and Robert, 1994;Richardson and Green, 1997;McLachlan and Peel, 2004) and others that are based on relabelling algorithms (Celeux, 1998;Stephens, 2000b;Rodriguez and Walker, 2014;Celeux et al, 2000). For a review and comparison of these methods see, for example, Jasra et al (2005) and Sperrin et al (2010).…”

Section: Introductionmentioning

confidence: 99%

Bayesian analysis for mixtures of discrete distributions with a non-parametric component

Alhaji

Dai

Hayashi

et al. 2015

Journal of Applied Statistics

View full text Add to dashboard Cite

Bayesian finite mixture modelling is a flexible parametric modelling approach for classification and density fitting. Many application areas require distinguishing a signal from a noise component. In practice, it is often difficult to justify a specific distribution for the signal component, therefore the signal distribution is usually further modelled via a mixture of distributions. However, modelling the signal as a mixture of distributions is computationally challenging due to the difficulties in justifying the exact number of components to be used and due to the label switching problem. This paper proposes the use of a non-parametric distribution to model the signal component. We consider the case of discrete data and show how this new methodology leads to more accurate parameter estimation and smaller classification error. Moreover, it does not incur the label switching problem. We show an application of the method to data generated by ChIP-sequencing experiments.

show abstract

“…Many methods exist on how to tackle the label switching problem, for example, imposing identifiability constraints (Diebolt and Robert (1994), Richardson and Green (1997), McLachlan and Peel (2004)) and other methods based on relabelling algorithms (Celeux (1998), Stephens (2000b), Celeux et al (2000), Rodriguez and Walker (2014)). For a review and comparison of these methods see, for example, Jasra et al (2005) and Sperrin et al (2010).…”

Section: Introductionmentioning

confidence: 99%

Analysis of ChIP-seq Data Via Bayesian Finite Mixture Models with a Non-parametric Component

Alhaji

Dai

Hayashi

et al. 2016

Analysis of Large and Complex Data

View full text Add to dashboard Cite

Abstract. In large discrete data sets which requires classification into signal and noise components, the distribution of the signal is often very bumpy and does not follow a standard distribution. Therefore the signal distribution is further modelled as a mixture of component distributions.However, when the signal component is modelled as a mixture of distributions, we are faced with the challenges of justifying the number of components and the label switching problem (caused by multi-modality of the likelihood function). To circumvent these challenges, we propose a non-parametric structure for the signal component. This new method is more efficient in terms of precise estimates and better classifications. We demonstrate the efficacy of the methodology using a ChIPsequencing data set.

show abstract

Label Switching in Bayesian Mixture Models: Deterministic Relabeling Strategies

Abstract: In the folder "LabelSwitching" we have included all the code used in Subsection: Relabeling algorithms, Gibbs Sampler. This will enable readers to replicate all the analysis, simulations and

Cited by 77 publications

References 19 publications

label.switching: AnRPackage for Dealing with the Label Switching Problem in MCMC Outputs

label.switching: AnRPackage for Dealing with the Label Switching Problem in MCMC Outputs

Bayesian analysis for mixtures of discrete distributions with a non-parametric component

Analysis of ChIP-seq Data Via Bayesian Finite Mixture Models with a Non-parametric Component

Contact Info

Product

Resources

About