A Nearest Neighbors Quadrature for Posterior Approximation via Adaptive Sequential Design

Fernando Llorente; Luca Martino; Victor Elvira; David Delgado

doi:10.1109/SSP49050.2021.9513802

A Nearest Neighbors Quadrature for Posterior Approximation via Adaptive Sequential Design

Fernando Llorente, Luca Martino, Victor Elvira, David Delgado

School of Mathematics

Research output: Contribution to conference › Paper › peer-review

Abstract

We introduce a novel adaptive quadrature scheme based on a Nearest Neighbors (NN) approach and a sequential design procedure. The nodes of the quadrature are adaptively chosen by maximizing a suitable acquisition function. The proposed method is a powerful tool for the integration and emulation of complex posterior distributions. Numerical results show the advantage of the proposed approach with respect to Markov Chain Monte Carlo (MCMC) and importance sampling algorithms.

Original language	English
Pages	301-305
Number of pages	5
DOIs	https://doi.org/10.1109/SSP49050.2021.9513802
Publication status	Published - 21 Aug 2021
Event	21st IEEE Statistical Signal Processing Workshop, SSP 2021 - Virtual, Rio de Janeiro, Brazil Duration: 11 Jul 2021 → 14 Jul 2021

Conference

Conference	21st IEEE Statistical Signal Processing Workshop, SSP 2021
Country/Territory	Brazil
City	Virtual, Rio de Janeiro
Period	11/07/21 → 14/07/21

Keywords / Materials (for Non-textual outputs)

active learning
Bayesian inference
Bayesian quadrature
MCMC
sequential design

Access to Document

10.1109/SSP49050.2021.9513802

Cite this

@conference{0b02e1e517a942e78b4853d17dafcf6f,

title = "A Nearest Neighbors Quadrature for Posterior Approximation via Adaptive Sequential Design",

abstract = "We introduce a novel adaptive quadrature scheme based on a Nearest Neighbors (NN) approach and a sequential design procedure. The nodes of the quadrature are adaptively chosen by maximizing a suitable acquisition function. The proposed method is a powerful tool for the integration and emulation of complex posterior distributions. Numerical results show the advantage of the proposed approach with respect to Markov Chain Monte Carlo (MCMC) and importance sampling algorithms.",

keywords = "active learning, Bayesian inference, Bayesian quadrature, MCMC, sequential design",

author = "Fernando Llorente and Luca Martino and Victor Elvira and David Delgado",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 21st IEEE Statistical Signal Processing Workshop, SSP 2021 ; Conference date: 11-07-2021 Through 14-07-2021",

year = "2021",

month = aug,

day = "21",

doi = "10.1109/SSP49050.2021.9513802",

language = "English",

pages = "301--305",

}

TY - CONF

T1 - A Nearest Neighbors Quadrature for Posterior Approximation via Adaptive Sequential Design

AU - Llorente, Fernando

AU - Martino, Luca

AU - Elvira, Victor

AU - Delgado, David

PY - 2021/8/21

Y1 - 2021/8/21

N2 - We introduce a novel adaptive quadrature scheme based on a Nearest Neighbors (NN) approach and a sequential design procedure. The nodes of the quadrature are adaptively chosen by maximizing a suitable acquisition function. The proposed method is a powerful tool for the integration and emulation of complex posterior distributions. Numerical results show the advantage of the proposed approach with respect to Markov Chain Monte Carlo (MCMC) and importance sampling algorithms.

AB - We introduce a novel adaptive quadrature scheme based on a Nearest Neighbors (NN) approach and a sequential design procedure. The nodes of the quadrature are adaptively chosen by maximizing a suitable acquisition function. The proposed method is a powerful tool for the integration and emulation of complex posterior distributions. Numerical results show the advantage of the proposed approach with respect to Markov Chain Monte Carlo (MCMC) and importance sampling algorithms.

KW - active learning

KW - Bayesian inference

KW - Bayesian quadrature

KW - MCMC

KW - sequential design

UR - https://www.scopus.com/pages/publications/85113419637

U2 - 10.1109/SSP49050.2021.9513802

DO - 10.1109/SSP49050.2021.9513802

M3 - Paper

AN - SCOPUS:85113419637

SP - 301

EP - 305

T2 - 21st IEEE Statistical Signal Processing Workshop, SSP 2021

Y2 - 11 July 2021 through 14 July 2021

ER -

A Nearest Neighbors Quadrature for Posterior Approximation via Adaptive Sequential Design

Abstract

Conference

Keywords / Materials (for Non-textual outputs)

Access to Document

Other files and links

Fingerprint

Cite this