Skip to main content Skip to navigation

2013-2014

2013/2014 Term 3:

07/03/2014: Louis Aslett (Oxford; i-like), Coupled Hidden Markov Models
11/04/2014: Andreas Hetland

2013/2014 Term 2:

14/02/2014: Paul Jenkins,
Exact simulation of the sample paths of a diffusion with a finite entrance boundary, mostly based on this paper plus some extensions.
31/01/2014: Jere Koskela,
will present the paper: The time machine: a simulation approach for stochastic trees
24/01/2014: Sergios Agapiou,
will present the paper: Analysis of the Gibbs sampler for hierarchical inverse problems
17/01/2014: Christopher Drovandi (Queensland University of Technology/Warwick)
Bayesian Indirect Inference
Indirect inference (II) is a methodology for estimating the parameters of an intractable (generative) model on the basis of an alternative (auxiliary) model that is both analytically and computationally easier to deal with. Such an approach has been well explored in the classical literature but has received substantially less attention in the Bayesian paradigm. In this talk I will compare and contrast a collection of so-called Bayesian indirect inference (BII) methods. One class of BII methods uses approximate Bayesian computation (referred to here as ABC II) where the summary statistic is formed on the basis of the auxiliary model, using ideas from II. Another approach proposed in the literature, referred to here as Bayesian indirect likelihood (BIL), is a fundamentally different approach to ABC II. New theoretical results for BIL are devised to give extra insights into its approximation behaviour and also its differences with ABC II. The results, insights and comparisons are illustrated on some toy and more substantive applications. One application involves estimating the parameters of a trivariate stochastic process describing the evolution of macroparasites within a host based on real data.

2013/2014 Term 1:

06/12/2013: Axel Finke (University of Warwick)

Static-parameter estimation in piecewise deterministic processes using particle Gibbs samplers

We give a brief introduction to piecewise deterministic processes (PDPs). The latter form a class of stochastic processes that jump randomly at a countable number of stopping times but otherwise evolve deterministically in continuous time.
We develop a particle Gibbs sampler for static-parameter estimation in PDPs that are observed only partially, noisily and in discrete time. We present a reformulation of the original particle filter for PDPs. This permits the use of a variance-reduction technique known as ancestor sampling that greatly improves mixing of the particle Gibbs chain.

This is joint work with Adam Johansen and Dario Spanò.
29/11/2013: Christopher Drovandi (Queensland University of Technology/Warwick)
15/11/2013: Daniel Rudolf (University of Jena, Germany)
Title: On the hybrid slice sampler
01/11/2013: Ingeborg Waernbaum (Department of Statistics, Umeå University, Sweden)
Title:
Data-driven Algorithms for Dimension Reduction in Causal Inference: analyzing the effect of school achievements on acute complications of type 1 diabetes mellitus
Abstract:
In observational studies, the causal effect of a treatment may be confounded with variables that are related to both the treatment and the outcome of interest. In order to identify a causal effect, such studies often rely on the unconfoundedness assumption, i.e., that all confounding variables are observed. The choice of covariates to control for, which is primarily based on subject matter knowledge, may result in a large covariate vector in the attempt to ensure that unconfoundedness holds. However, including redundant covariates is suboptimal when the effect is estimated nonparametrically, e.g., due to the curse of dimensionality. In this paper, datadriven algorithms for the selection of suffcient covariate subsets are investigated. Under the assumption of unconfoundedness we search for minimal subsets of the covariate vector. Based on the framework of suffcient dimension reduction or kernel smoothing, the algorithms perform a backward elimination procedure testing the significance of each covariate. Their performance is evaluated in simulations and an application using data from the Swedish Childhood Diabetes Register is also presented.
Joint with : Emma Persson, Jenny Häggström and Xavier de Luna.