![]() |
Royal Statistical Society |
2000 | |
---|---|
5.00 p.m., 18 October |
OLE BARNDORFF-NIELSEN (University of Aarhus) and
NEIL SHEPHARD (Nuffield College, Oxford)
Non-Gaussian OU based models and some of their uses in financial economics Non-Gaussian processes of Ornstein-Uhlenbeck (OU) type, and combinations of such and with Brownian motion, offer the possibility of capturing distributional deviations from Gaussianity and for flexible modelling of dependence structures. Their power is illustrated by a sustained application within the context of finance and econometrics. We construct continuous time stochastic volatility models and we study these models in relation to financial data and theory. Appears (with discussion) in Journal of the Royal Statistical Society, Series B, 63(2), 167-241 (2001). |
5.00 p.m., 13 December |
MARC KENNEDY (NISS) and
TONY O'HAGAN (University of Sheffield)
Bayesian Calibration of Computer Models The importance of quantifying uncertainty in computer models is increasingly recognised by users and developers of models. Often, data are available with which to calibrate the model, i.e. to learn about those unknown parameters which define the context in which the model is to be used. We present a Bayesian approach in which residual uncertainty after calibration is recognised. Our model also offers the possibility of 'correcting' model deficiencies. Appears (with discussion) in Journal of the Royal Statistical Society, Series B, 63(3), 425-464 (2001). |
2001 | |
5.00 p.m., 14 February |
CHRIS GLASBEY (BioSS) and
KANTI MARDIA (University of Leeds)
A penalised likelihood approach to image warping Warping functions, which deform images by mapping between image domains, are a key component of imaging technology. We estimate these functions by maximising a penalised likelihood, strategically constructed through a new image model to measure similarity between images and new distortion criteria to penalise warpings. The power of the method is illustrated through registering a remotely-sensed image, aligning microscope images, and discriminating between species of fish. Appears (with discussion) in Journal of the Royal Statistical Society, Series B, 63(3), 465-514 (2001). |
5.00 p.m., 23 May |
SUE LEWIS (University of Southampton) and
ANGELA DEAN (Ohio State University)
Detection of interactions in experiments on large numbers of factors Valuable information on interactions in factorial experiments is often sacrificed to achieve economical sized experiments. Group screening strategies, with particular emphasis on the detection of key interactions, will be discussed for a large number of factors. Methodology is developed for guiding choice of screening technique, and choosing sizes of factor groupings. Examples will illustrate the methodology and issues for practical implementation. Appears (with discussion) in Journal of the Royal Statistical Society, Series B, 63(4), 633-672 (2001). |