Independent Doubly Adaptive Rejection Metropolis Sampling Within Gibbs Sampling

Bayesian methods have become very popular in signal processing lately, even though performing exact Bayesian inference is often unfeasible due to the lack of analytical expressions for optimal Bayesian estimators. In order to overcome this problem, Monte Carlo (MC) techniques are frequently used. Several classes of MC schemes have been developed, including Markov Chain Monte Carlo (MCMC) methods, particle filters and population Monte Carlo approaches. In this paper, we concentrate on the Gibbs-type approach, where automatic and fast samplers are needed to draw from univariate (full-conditional) densities. The Adaptive Rejection Metropolis Sampling (ARMS) technique is widely used within Gibbs sampling, but suffers from an important drawback: an incomplete adaptation of the proposal in some cases. In this work, we propose an alternative adaptive MCMC algorithm (IA2RMS) that overcomes this limitation, speeding up the convergence of the chain to the target, allowing us to simplify the construction of the sequence of proposals, and thus reducing the computational cost of the entire algorithm. Note that, although IA2RMS has been developed as an extremely efficient MCMC-within-Gibbs sampler, it also provides an excellent performance as a stand-alone algorithm when sampling from univariate distributions. In this case, the convergence of the proposal to the target is proved and a bound on the complexity of the proposal is provided. Numerical results, both for univariate (stand-alone IA2RMS) and multivariate (IA2RMS-within-Gibbs) distributions, show that IA2RMS outperforms ARMS and other classical techniques, providing a correlation among samples close to zero.

[1]  By W. R. GILKSt,et al.  Adaptive Rejection Sampling for Gibbs Sampling , 2010 .

[2]  Mónica F. Bugallo,et al.  Joint Model Selection and Parameter Estimation by Population Monte Carlo Simulation , 2010, IEEE Journal of Selected Topics in Signal Processing.

[3]  Karl-Rudolf Koch,et al.  Gibbs sampler by sampling-importance-resampling , 2007 .

[4]  Colin Fox,et al.  A Gibbs sampler for conductivity imaging and other inverse problems , 2012, Other Conferences.

[5]  Renate Meyer,et al.  Metropolis–Hastings algorithms with adaptive proposals , 2008, Stat. Comput..

[6]  Andreas M. Ali,et al.  An Empirical Study of Collaborative Acoustic Source Localization , 2007, 2007 6th International Symposium on Information Processing in Sensor Networks.

[7]  Chein-I Chang,et al.  Semi-Supervised Linear Spectral Unmixing Using a Hierarchical Bayesian Model for Hyperspectral Imagery , 2008, IEEE Transactions on Signal Processing.

[8]  Luca Martino,et al.  Independent doubly Adaptive Rejection Metropolis Sampling , 2014, 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[9]  Wolfgang Hörmann,et al.  A rejection technique for sampling from T-concave distributions , 1995, TOMS.

[10]  J. Besag,et al.  Spatial Statistics and Bayesian Computation , 1993 .

[11]  J. Gåsemyr.  On an adaptive version of the Metropolis-Hastings algorithm with independent proposal distribution , 2003 .

[12]  William J. Fitzgerald,et al.  Markov chain Monte Carlo methods with applications to signal processing , 2001, Signal Process..

[13]  Mónica F. Bugallo,et al.  A sequential Monte Carlo technique for blind synchronization and detection in frequency-flat Rayleigh fading wireless channels , 2004, Signal Process..

[14]  Wolfgang Hörmann,et al.  A sweep-plane algorithm for generating random tuples in simple polytopes , 1998, Math. Comput..

[15]  L. Holden,et al.  Adaptive independent Metropolis–Hastings , 2009, 0903.0483.

[16]  Heikki Haario,et al.  DRAM: Efficient adaptive MCMC , 2006, Stat. Comput..

[17]  Jean-Yves Tourneret,et al.  Classification of chirp signals using hierarchical Bayesian learning and MCMC methods , 2002, IEEE Trans. Signal Process..

[18]  R. Carroll,et al.  Advanced Markov Chain Monte Carlo Methods: Learning from Past Samples , 2010 .

[19]  L. Tierney.  Exploring Posterior Distributions Using Markov Chains , 1992 .

[20]  M. Pitt,et al.  Filtering via Simulation: Auxiliary Particle Filters , 1999 .

[21]  W. Gilks,et al.  Adaptive Rejection Metropolis Sampling Within Gibbs Sampling , 1995 .

[22]  W. K. Hastings,et al.  Monte Carlo Sampling Methods Using Markov Chains and Their Applications , 1970 .

[23]  Wei Shao,et al.  An efficient proposal distribution for Metropolis-Hastings using a B-splines technique , 2013, Comput. Stat. Data Anal..

[24]  S. Godsill,et al.  Special issue on Monte Carlo methods for statistical signal processing , 2002 .

[25]  Renate Meyer,et al.  Bayesian semiparametric modeling of survival data based on mixtures of B-spline distributions , 2011, Comput. Stat. Data Anal..

[26]  Christophe Andrieu,et al.  A tutorial on adaptive MCMC , 2008, Stat. Comput..

[27]  H. Haario,et al.  An adaptive Metropolis algorithm , 2001 .

[28]  Kevin P. Murphy,et al.  Machine learning - a probabilistic perspective , 2012, Adaptive computation and machine learning series.

[29]  Xiaodong Wang,et al.  Monte Carlo methods for signal processing , 2005 .

[30]  Roman Karawatzki.  The Multivariate Ahrens Sampling Method , 2006 .

[31]  George Y. Sofronov,et al.  Adaptive independence samplers , 2008, Stat. Comput..

[32]  Steve P. Brooks,et al.  Output Assessment for Monte Carlo Simulations via the Score Statistic , 2006 .

[33]  Luca Martino,et al.  Generalized rejection sampling schemes and applications in signal processing , 2009, Signal Process..

[34]  Robert L. Smith,et al.  Hit-and-Run Algorithms for Generating Multivariate Distributions , 1993, Math. Oper. Res..

[35]  N. Metropolis,et al.  The Monte Carlo method. , 1949 .

[36]  J Besag,et al.  DISCUSSION ON THE MEETING ON THE GIBBS SAMPLER AND OTHER MARKOV CHAIN-MONTE CARLO METHODS , 1993 .

[37]  Nizar Bouguila,et al.  Bayesian learning of finite generalized Gaussian mixture models on images , 2011, Signal Process..

[38]  K. Riedel.  Numerical Bayesian Methods Applied to Signal Processing , 1996 .

[39]  Jakob Creutzig,et al.  W. Hörmann, J. Leydold, G. Derflinger: Automatic nonuniform random variate generation , 2006 .

[40]  James P. Reilly,et al.  Reversible jump MCMC for joint detection and estimation of sources in colored noise , 2002, IEEE Trans. Signal Process..

[41]  Wolfgang Hörmann,et al.  Automatic Nonuniform Random Variate Generation , 2011 .

[42]  Petar M. Djuric,et al.  Perfect sampling: a review and applications to signal processing , 2002, IEEE Trans. Signal Process..

[43]  Heikki Haario,et al.  Adaptive proposal distribution for random walk Metropolis algorithm , 1999, Comput. Stat..

[44]  Petar M. Djuric,et al.  Gibbs sampling approach for generation of truncated multivariate Gaussian random variables , 1999, 1999 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings. ICASSP99 (Cat. No.99CH36258).

[45]  Nando de Freitas,et al.  An Introduction to MCMC for Machine Learning , 2004, Machine Learning.

[46]  Ming-Hui Chen,et al.  Toward Black-Box Sampling: A Random-Direction Interior-Point Markov Chain Approach , 1998 .

[47]  G. Warnes.  The Normal Kernel Coupler: An Adaptive Markov Chain Monte Carlo Method for Efficiently Sampling From Multi-Modal Distributions , 2001 .

[48]  Lars Holden,et al.  Adaptive Chains , 1998 .

[49]  Walter R. Gilks,et al.  Adaptive Direction Sampling , 1994 .

[50]  K OrJ.  Numerical Bayesian methods applied to signal processing , 1996 .

[51]  Walter R. Gilks,et al.  Corrigendum: Adaptive Rejection Metropolis Sampling , 1997 .

[52]  Luca Martino,et al.  On the flexibility of the design of multiple try Metropolis schemes , 2012, Computational Statistics.

[53]  Luca Martino,et al.  A generalization of the adaptive rejection sampling algorithm , 2010, Stat. Comput..

[54]  J. Rosenthal,et al.  Coupling and Ergodicity of Adaptive Markov Chain Monte Carlo Algorithms , 2007, Journal of Applied Probability.

[55]  Peter Jaeckel,et al.  Monte Carlo methods in finance , 2002 .

[56]  Renate Meyer,et al.  Adaptive rejection Metropolis sampling using Lagrange interpolation polynomials of degree 2 , 2008, Comput. Stat. Data Anal..

[57]  Dani Gamerman,et al.  Markov Chain Monte Carlo: Stochastic Simulation for Bayesian Inference , 1997 .

[58]  L. Tierney.  Markov Chains for Exploring Posterior Distributions , 1994 .

[59]  Josef Leydold,et al.  A rejection technique for sampling from log-concave multivariate distributions , 1998, TOMC.

[60]  Hoon Kim,et al.  Monte Carlo Statistical Methods , 2000, Technometrics.

[61]  H Putter,et al.  A Bayesian approach to parameter estimation in HIV dynamical models , 2002, Statistics in medicine.

[62]  M. Evans,et al.  Random Variable Generation Using Concavity Properties of Transformed Densities , 1998 .

[63]  Alejandro Jara,et al.  The Polya Tree Sampler: Toward Efficient and Automatic Independent Metropolis–Hastings Proposals , 2011, Journal of computational and graphical statistics : a joint publication of American Statistical Association, Institute of Mathematical Statistics, Interface Foundation of North America.

[64]  Tim Hesterberg,et al.  Monte Carlo Strategies in Scientific Computing , 2002, Technometrics.

[65]  N. Metropolis,et al.  Equation of State Calculations by Fast Computing Machines , 1953, Resonance.

[66]  Y. Teh,et al.  Concave-Convex Adaptive Rejection Sampling , 2011 .

[67]  L. Devroye.  Non-Uniform Random Variate Generation , 1986 .

[68]  Neil J. Gordon,et al.  Editors: Sequential Monte Carlo Methods in Practice , 2001 .

[69]  Stephen G. Walker,et al.  On adaptive Metropolis–Hastings methods , 2013, Stat. Comput..

[70]  Nicholas G. Polson,et al.  Particle Filtering , 2006 .

[71]  Xiaodong Wang,et al.  Monte Carlo methods for signal processing: a review in the statistical signal processing context , 2005, IEEE Signal Processing Magazine.

[72]  C. Andrieu,et al.  On the ergodicity properties of some adaptive MCMC algorithms , 2006, math/0610317.

[73]  Maurizio Dapor.  Monte Carlo Strategies , 2020, Transport of Energetic Electrons in Solids.