Bayesian Inference for Big Data with Stochastic Gradient Markov Chain Monte Carlo

Lead Research Organisation: University of Oxford

Department Name: Statistics

Abstract

Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.

Funded Value:

£158,970

Funded Period:

Jul 13 - Sep 16

Funder:

EPSRC

Project Status:

Closed

Project Category:

Research Grant

Project Reference:

EP/K009850/1

Principal Investigator:

Arnaud Doucet

Research Subject:

Info. & commun. Technol. (50%)

Mathematical sciences (50%)

Research Topic:

Artificial Intelligence (50%)

Statistics & Appl. Probability (50%)

Organisations

University of Oxford (Lead Research Organisation)

People	ORCID iD
Arnaud Doucet (Principal Investigator)

Publications

Author Name Title Publication

Date Published

|< < 1 2 > >|

10 25 50

Agapiou S (2014) Unbiased Monte Carlo: posterior estimation for intractable/infinite-dimensional models

Teh, YW (2015) Distributed Bayesian Learning with Stochastic Natural-gradient Expectation Propagation and the Posterior Server

Finke A (2016) On embedded hidden Markov models and particle Markov chain Monte Carlo methods

Yildirim S (2018) Scalable Monte Carlo inference for state-space models

A Bouchard-Côté (2015) The Bouncy Particle Sampler: A Non-Reversible Rejection-Free Markov Chain Monte Carlo Method

Vollmer S (2015) (Non-) asymptotic properties of Stochastic Gradient Langevin Dynamics

Alenlöv J (2016) Pseudo-Marginal Hamiltonian Monte Carlo

Bouchard-Côté A (2015) Particle Gibbs Split-Merge Sampling for Bayesian Inference in Mixture Models

Hasenclever L (2015) Distributed Bayesian Learning with Stochastic Natural-gradient Expectation Propagation and the Posterior Server

Döbler, C (2015) An iterative technique for bounding derivatives of solutions of Stein equations

Teh Y (2014) Consistency and fluctuations for stochastic gradient Langevin dynamics

Key Findings


Description	We are studying sophisticated new statistical methods to analyze big data sets. Current methods are very computationally intensive and do not scale in presence of big data. We are developing scalable yet sophisticated techniques to extract useful information from massive datasets.
Exploitation Route	There is still a lot of room for improvement, both methodologically and theoretically. So we expect over the forthcoming year to develop further our new algorithms.
Sectors	Aerospace Defence and Marine Digital/Communication/Information Technologies (including Software) Electronics Security and Diplomacy
URL	http://www.stats.ox.ac.uk/~doucet/journalsbysubject.html

Abstract

Organisations

People

ORCID iD

Publications