This paper reviews a technique of adaptive wavelet expansions and introduces the novel
concept of “biased wavelets.” These are functions that are localized in time and in frequency
but, unlike conventional wavelets, have an adjustable nonzero mean component. Under
mild conditions, it is shown that a conventional mother wavelet can be used to construct a
family of biased wavelets which spans the set of finite-energy functions L2 (R). Numerical
tests suggest that the introduction of the adjustable “bias” considerably improves the …