De-Noising By Soft-Thresholding-英文文献
DE-NOISING BY
SOFT-THRESHOLDING
David L. Donoho
Department of Statistics
Stanford University
Abstract
Donoho and Johnstone (1992a) prop osed a metho d for reconstruct-
ing an unknown function f on [0 ; 1] from noisy data d = f (t ) + z ,
i i i
iid
^
i = 0 ; : : : ; n 1, t = i=n , z N (0 ; 1). The reconstruction f is
i i n
dened in the wavelet domain by translating all the empirical wavelet
p p
co ecients of d towards 0 by an amount 2 log(n ) = n . We prove
^
two results ab out that estimator. [Smo oth]: With high probability f
n
is at least as smo oth as f , in any of a wide variety of smo othness mea-
sures. [Adapt]: The estimator comes nearly as close in mean square
to f as any measurable estimator can come, uniformly over balls in
each of two broad scales of smo othness classes. These two prop erties
are unprecedented in several ways. Our pro of of these results develops
new facts ab out abstract statistical inference and its connection with