Estimating fMRI Time Series Models in the Wavelet Domain

International audienceThis work addresses two main problems in wavelet-based time series estimation. The first problem is ML estimation of regression model parameters in the context of long-memory errors; as well as MLE of long-memory parameters characterising the noise structure. This method exploits the whitening properties of wavelet bases. In the second part, we consider a regularized estimator of the non-parametric component in partially linear models using wavelet expansions. Some key results are presented and illustrated using simulation and real functional MRI datasets