Model selection for the robust efficient signal processing observed with small L\'evy noise

TL;DR

This paper introduces a new model selection method for robust nonparametric signal estimation in the presence of impulse noise modeled by non-Gaussian Lévy processes, providing sharp oracle inequalities and asymptotic efficiency.

Contribution

It develops the first non-asymptotic sharp oracle inequalities for such models and demonstrates the asymptotic efficiency of the proposed estimators.

Findings

Established non-asymptotic sharp oracle inequalities for quadratic and robust risks.

Proved asymptotic efficiency of the estimators in high signal-to-noise ratio regimes.

Applied the methods to multi-path information transmission signal detection.

Abstract

We develop a new model selection method for the adaptive robust efficient nonparametric signal estimation observed with impulse noise which is defined by the general non Gaussian L\'evy processes. On the basis of the developed method, we construct the estimation procedures which are analyzed in two settings: in non asymptotic and asymptotic ones. For the first time for such models we show non asymptotic sharp oracle inequalities for the quadratic and for the robust risks, i.e. we show that the constructed procedures are optimal in the sharp oracle inequalities sense. Next, by making use of the obtained oracle inequalities, we provide the asymptotic efficiency property for the developed estimation methods in the adaptive setting when the signal/noise ratio goes to infinity. We apply the developed model selection methods for the signals number detection problem in multi-path information transmission.