DATE AND TIME:     Monday, September 29, 1997, 4:10 p.m.

PLACE:             319 Snedecor Hall

SPEAKER:           Hsin-Cheng Huang
                   Department of Statistics
                   Iowa State University
                   
TITLE:             Deterministic/Stochastic Wavelet Decomposition for
                   Recovery of Noisy Data

                       ABSTRACT

    In a series of papers on nonparametric regression, D. Donoho and I.
Johnstone develop wavelet shrinkage methods for recovering unknown
deterministic signals from noisy data. We propose a new rationale for wavelet
shrinkage, based on the assumption that the underlying process can be
decomposed into a deterministic trend plus a zero-mean dependent signal plus
noise. This assumption, which is common in spatial statistics and time series,
leads one to rethink the way spatial and temporal data should be analyzed
using wavelets. Our approach takes the dependencies of empirical wavelet
coefficients, both within scales and across scales, into account. Simulation
studies show that our method is at least as good as the universal-thresholding
and the SURE-thresholding methods when the signals are purely deterministic.
It outperforms those thresholding methods on signals that contain 
certain stochastic components.

COFFEE: 3:45 p.m., 104 Snedecor Hall