Asymptotic Equivalence of Density Estimation and Gaussian White Noise

Signal recovery in Gaussian white noise with variance tending to zero has served for some time as a representative model for nonparametric curve estimation, having all the essential traits in a pure form. The equivalence has mostly been stated informally, but an approximation in the sense of Le Cam's deficiency distance Δ would make it precise. The models are then asymptotically equivalent for all purposes of statistical decision with bounded loss. In nonparametrics, a first result of this kind has recently been established for Gaussian regression. We consider the analogous problem for the experiment given by n i.i.d. observations having density f on the unit interval. Our basic result concerns the parameter space of densities which are in a Holder ball with exponent α > 1/2 and which are uniformly bounded away from zero. We show that an i. i. d. sample of size n with density f is globally asymptotically equivalent to a white noise experiment with drift f l/2 and variance 1/4n -l . This represents a nonparametric analog of Le Cam's heteroscedastic Gaussian approximation in the finite dimensional case. The proof utilizes empirical process techniques related to the Hungarian construction. White noise models on f and log f are also considered, allowing for various automatic asymptotic risk bounds in the i.i.d. model from white noise.

[1]  Michael Woodroofe,et al.  On the Maximum Deviation of the Sample Density , 1967 .

[2]  M. Nikolskii,et al.  Approximation of Functions of Several Variables and Embedding Theorems , 1971 .

[3]  K. Parthasarathy.  Introduction to Probability and Measure , 1979 .

[4]  P. Millar.  Asymptotic minimax theorems for the sample distribution function , 1979 .

[5]  Lucien Birgé.  Approximation dans les espaces métriques et théorie de l'estimation , 1983 .

[6]  M. Nussbaum.  Spline Smoothing in Regression Models and Asymptotic Efficiency in $L_2$ , 1985 .

[7]  Lucien Le Cam,et al.  Sur l'approximation de familles de mesures par des familles gaussiennes , 1985 .

[8]  James V. Bondar,et al.  Mathematical theory of statistics , 1985 .

[9]  Enno Mammen,et al.  The Statistical Information Contained in Additional Observations , 1986 .

[10]  L. L. Cam,et al.  Asymptotic Methods In Statistical Decision Theory , 1986 .

[11]  J. Wellner,et al.  Empirical Processes with Applications to Statistics , 2009 .

[12]  S. Geer.  Estimating a Regression Function , 1990 .

[13]  R. M. Dudley,et al.  Real Analysis and Probability , 1989 .

[14]  M. Nussbaum,et al.  A Risk Bound in Sobolev Class Regression , 1990 .

[15]  Grace L. Yang,et al.  Asymptotics In Statistics , 1990 .

[16]  D. Donoho,et al.  Minimax Risk Over Hyperrectangles, and Implications , 1990 .

[17]  Renormalization and White Noise Approximation for Nonparametric Functional Estimation Problems , 1992 .

[18]  Rolf-Dieter Reiss,et al.  A Course on Point Processes , 1992 .

[19]  D. Donoho,et al.  Renormalization Exponents and Optimal Pointwise Rates of Convergence , 1992 .

[20]  Poisson approximation of empirical processes , 1992 .

[21]  LAN in Problems of Nonparametric Estimation of Functions and Lower Bounds for Quadratic Risks , 1992 .

[22]  Mark G. Low.  Renormalizing Upper and Lower Bounds for Integrated Risk in the White Noise Model , 1993 .

[23]  Emmanuel Rio,et al.  Local invariance principles and their application to density estimation , 1994 .

[24]  Vladimir Koltchinskii,et al.  Komlos-Major-Tusnady approximation for the general empirical process and Haar expansions of classes of functions , 1994 .

[25]  A. Korostelev.  An Asymptotically Minimax Regression Estimator in the Uniform Norm up to Exact Constant , 1994 .

[26]  I. Johnstone,et al.  Wavelet Shrinkage: Asymptopia? , 1995 .

[27]  L. Brown,et al.  Asymptotic equivalence of nonparametric regression and white noise , 1996 .