\documentclass[12pt]{article}
\usepackage{amsmath,amssymb,amsfonts}
\begin{document}
This work deals with forecasting time series using wavelets and ker�nel smoothing. A forecasting procedure can be defined by estimating the prediction equation by direct regression of the process on the non�decimated wavelet coefficients depending on its past values. In the same context, after the seminal work of Renaud et al. [RSM03], we study a generalization of the prediction procedure associating kernel smoothing and wavelets. We then illustrate the proposed procedure on non-stationary simulated and real data.
\end{document}