| Prof. Marc Raimondo
(
School of Mathematics and Statistics,
The University of Sydney )
Large Deviations, Extreme Value Theory and Wavelet Transforms.
摘要
| | A key distribution in wavelet statistical methods is that of the maximum of (some function of) noisy wavelet transforms. Indeed, existing wavelet thresholding methods are obtained using asymptotic theory for the maximum of Gaussian processes. Recent papers propose alternative methods to model extreme values of noisy wavelet transforms. The approach is based on the so-called Peaks Over Threshold models. The idea is that excesses over high threshold follow a General Pareto Distribution (GPD). In this talk, I will review properties of the GPD-fit to noisy wavelet transforms. |
91年7月5日 (星期五)
下午4:00-5:00
台灣大學數學系新數館308室
|