Non-parametric regression with observations at imprecise times: Bayesian radiocarbon calibration
Add to your list(s)
Download to your calendar using vCal
If you have a question about this talk, please contact Richard Nickl.
For accurate radiocarbon dating, it is necessary to identify fluctuations in
the level of radioactive carbon (14C) over time. The processes that give
rise to these fluctuations are not understood and so a datum-based
calibration curve is required. Historic material can provide us with
observations of radiocarbon levels at distinct times in the past, but in
many cases the precise age of this material is unknown and must itself be
estimated.
With some exceptions, the current non-parametric regression literature
concentrates on the case where the times at which we observe the function
are known precisely. In this talk we will present a general methodology for
non-parametric regression, developed as part of this on-going radiocarbon
calibration work, which can handle such imprecise time estimates.
In our approach, the underlying function is modelled as a Weiner process
with sample paths that are updated, in light of the imprecise data, through
use of a Metropolis-within-Gibbs algorithm. This methodology is
computationally efficient due to the independent increment property of the
process and use of a Brownian bridge.
http://maths.dept.shef.ac.uk/pas/staff_info.php?id=289
This talk is part of the Statistics series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
|