Group metropolis sampling

Luca Martino; Víctor Elvira; Gustau Camps-Valls

doi:10.23919/EUSIPCO.2017.8081197

Group metropolis sampling

Luca Martino, Víctor Elvira, Gustau Camps-Valls

School of Mathematics

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Monte Carlo (MC) methods are widely used for Bayesian inference and optimization in statistics, signal processing and machine learning. Two well-known class of MC methods are the Importance Sampling (IS) techniques and the Markov Chain Monte Carlo (MCMC) algorithms. In this work, we introduce the Group Importance Sampling (GIS) framework where different sets of weighted samples are properly summarized with one summary particle and one summary weight. GIS facilitates the design of novel efficient MC techniques. For instance, we present the Group Metropolis Sampling (GMS) algorithm which produces a Markov chain of sets of weighted samples. GMS in general outperforms other multiple try schemes as shown by means of numerical simulations.

Original language	English
Title of host publication	25th European Signal Processing Conference, EUSIPCO 2017
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	201-205
Number of pages	5
Volume	2017-January
ISBN (Electronic)	9780992862671
ISBN (Print)	978-1-5386-0751-0
DOIs	https://doi.org/10.23919/EUSIPCO.2017.8081197
Publication status	Published - 23 Oct 2017
Event	25th European Signal Processing Conference, EUSIPCO 2017 - Kos, Greece Duration: 28 Aug 2017 → 2 Sep 2017

Conference

Conference	25th European Signal Processing Conference, EUSIPCO 2017
Country/Territory	Greece
City	Kos
Period	28/08/17 → 2/09/17

Keywords

Bayesian inference
Gaussian processes (GP)
Importance Sampling
Markov Chain Monte Carlo (MCMC)

Access to Document

10.23919/EUSIPCO.2017.8081197

Cite this

@inproceedings{ad777df6da5449d3bc721973a62d502e,

title = "Group metropolis sampling",

abstract = "Monte Carlo (MC) methods are widely used for Bayesian inference and optimization in statistics, signal processing and machine learning. Two well-known class of MC methods are the Importance Sampling (IS) techniques and the Markov Chain Monte Carlo (MCMC) algorithms. In this work, we introduce the Group Importance Sampling (GIS) framework where different sets of weighted samples are properly summarized with one summary particle and one summary weight. GIS facilitates the design of novel efficient MC techniques. For instance, we present the Group Metropolis Sampling (GMS) algorithm which produces a Markov chain of sets of weighted samples. GMS in general outperforms other multiple try schemes as shown by means of numerical simulations.",

keywords = "Bayesian inference, Gaussian processes (GP), Importance Sampling, Markov Chain Monte Carlo (MCMC)",

author = "Luca Martino and V{\'i}ctor Elvira and Gustau Camps-Valls",

year = "2017",

month = oct,

day = "23",

doi = "10.23919/EUSIPCO.2017.8081197",

language = "English",

isbn = "978-1-5386-0751-0",

volume = "2017-January",

pages = "201--205",

booktitle = "25th European Signal Processing Conference, EUSIPCO 2017",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

address = "United States",

note = "25th European Signal Processing Conference, EUSIPCO 2017 ; Conference date: 28-08-2017 Through 02-09-2017",

}

TY - GEN

T1 - Group metropolis sampling

AU - Martino, Luca

AU - Elvira, Víctor

AU - Camps-Valls, Gustau

PY - 2017/10/23

Y1 - 2017/10/23

N2 - Monte Carlo (MC) methods are widely used for Bayesian inference and optimization in statistics, signal processing and machine learning. Two well-known class of MC methods are the Importance Sampling (IS) techniques and the Markov Chain Monte Carlo (MCMC) algorithms. In this work, we introduce the Group Importance Sampling (GIS) framework where different sets of weighted samples are properly summarized with one summary particle and one summary weight. GIS facilitates the design of novel efficient MC techniques. For instance, we present the Group Metropolis Sampling (GMS) algorithm which produces a Markov chain of sets of weighted samples. GMS in general outperforms other multiple try schemes as shown by means of numerical simulations.

AB - Monte Carlo (MC) methods are widely used for Bayesian inference and optimization in statistics, signal processing and machine learning. Two well-known class of MC methods are the Importance Sampling (IS) techniques and the Markov Chain Monte Carlo (MCMC) algorithms. In this work, we introduce the Group Importance Sampling (GIS) framework where different sets of weighted samples are properly summarized with one summary particle and one summary weight. GIS facilitates the design of novel efficient MC techniques. For instance, we present the Group Metropolis Sampling (GMS) algorithm which produces a Markov chain of sets of weighted samples. GMS in general outperforms other multiple try schemes as shown by means of numerical simulations.

KW - Bayesian inference

KW - Gaussian processes (GP)

KW - Importance Sampling

KW - Markov Chain Monte Carlo (MCMC)

UR - http://www.scopus.com/inward/record.url?scp=85041423929&partnerID=8YFLogxK

U2 - 10.23919/EUSIPCO.2017.8081197

DO - 10.23919/EUSIPCO.2017.8081197

M3 - Conference contribution

AN - SCOPUS:85041423929

SN - 978-1-5386-0751-0

VL - 2017-January

SP - 201

EP - 205

BT - 25th European Signal Processing Conference, EUSIPCO 2017

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 25th European Signal Processing Conference, EUSIPCO 2017

Y2 - 28 August 2017 through 2 September 2017

ER -

Group metropolis sampling

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this