In many fields of science the unit of experimental observation is a smooth scalar or vector continuum. Examples include: 1D time series, 2D images of the cosmos, and 3D functional brain image time series. Traditional analyses artificially discretize these continua and extract regional scalar/vector summary metrics, but this approach is often ad hoc and is rarely objective. In contrast, continuum statistics regard the measured continuum as non-simplifiable, and conduct hypothesis testing directly on a set of registered continuum observations, using a Random Field Theory (RFT) correction for multiple comparisons to retain a Type I error rate of alpha at the continuum level. This seminar will provide an overview of RFT and some of its common applications, with particular focus on the computations underlying RFT-based statistical inference, to emphasize how RFT can efficiently yet powerfully identify arbitrary n-dimensional continuum signals.
Last modified: Tuesday, 25-Feb-2014 10:52:38 NZDT
This page is maintained by the seminar list administrator.