Tue, 09 Nov 2010
16:00 -
17:00
DH 3rd floor SR
Tue, 26 Oct 2010
16:00 -
17:00
DH 3rd floor SR
Tue, 02 Nov 2010
16:00 -
17:00
DH 3rd floor SR
Geoghegan stacks and finiteness properties of groups
Benno Kuckuck
(Oxford University)
Abstract
Geoghegan's stack construction is a tool for analysing groups
that act on simply connected CW complexes, by providing a topological
description in terms of cell stabilisers and the quotient complex,
similar to what Bass-Serre theory does for group actions on trees. We
will introduce this construction and see how it can be used to give
results on finiteness properties of groups.
Tue, 23 Nov 2010
16:00 -
17:00
DH 3rd floor SR
Tue, 19 Oct 2010
16:00 -
17:00
DH 3rd floor SR
Tue, 12 Oct 2010
16:00 -
17:00
DH 3rd floor SR
Random Walks: Getting from A to A.
David Hume
(Oxford University)
Abstract
This talk introduces the topic of random walks on a finitely generated group and asks what properties of such a group can be detected through knowledge of such walks.
Thu, 07 May 2009
11:00
11:00
DH 3rd floor SR
Bayesian Gaussian Process models for multi-sensor time-series prediction
Michael Osborne
(Oxford University)
Abstract
We propose a powerful prediction algorithm built upon Gaussian
processes (GPs). They are particularly useful for their flexibility,
facilitating accurate prediction even in the absence of strong physical models. GPs further allow us to work within a completely Bayesian framework. As such, we show how the hyperparameters of our system can be marginalised by use of Bayesian Monte Carlo, a principled method of approximate integration. We employ the error bars of the GP's prediction as a means to select only the most informative observations to store. This allows us to introduce an iterative formulation of the GP to give a dynamic, on-line algorithm. We also show how our error bars can be used to perform active data selection, allowing the GP to select where and when it should next take a measurement.
We demonstrate how our methods can be applied to multi-sensor prediction problems where data may be missing, delayed and/or correlated. In particular, we present a real network of weather sensors as a testbed for our algorithm.
processes (GPs). They are particularly useful for their flexibility,
facilitating accurate prediction even in the absence of strong physical models. GPs further allow us to work within a completely Bayesian framework. As such, we show how the hyperparameters of our system can be marginalised by use of Bayesian Monte Carlo, a principled method of approximate integration. We employ the error bars of the GP's prediction as a means to select only the most informative observations to store. This allows us to introduce an iterative formulation of the GP to give a dynamic, on-line algorithm. We also show how our error bars can be used to perform active data selection, allowing the GP to select where and when it should next take a measurement.
We demonstrate how our methods can be applied to multi-sensor prediction problems where data may be missing, delayed and/or correlated. In particular, we present a real network of weather sensors as a testbed for our algorithm.
Thu, 12 Feb 2009
11:00
11:00
DH 3rd floor SR
Fri, 30 Jan 2009
16:30 -
17:00
DH 3rd floor SR
Modelling the Circulatory System: Evaluating Arterial Pressure and Cardiac Output
Athanasios Tsanas
(Oxford University)
Abstract
The circulatory system is the most important and amongst the most complicated mechanisms in the human body. Consisting of the heart, the arteries and the veins, it is amply aided by a variety of mechanisms aiming to facilitate adequate perfusion of the body tissues at the appropriate pressure. On this talk I am focusing on the development of a computational model which relates patient specific factors (age, gender, whether someone is an athlete/smokes etc) and their effects on different vascular regions which ultimately determine the arterial pressure and the cardiac output.
Thu, 26 Feb 2009
11:00
11:00
DH 3rd floor SR