Réduction de dimension par un nouvel estimateur de la distance de patrick Fisher é l'aide des fonctions orthogonales
Abstract
Here, we intend to introduce new estimators of the Patrick Fisher distance and the Euclidian probabilistic dependence measure by using orthogonal series. Therefore non parametric multiclass dimension reduction could be easily generalized to the multivariate case reduction. The performance of the proposed estimates will be studied by simulations in the case of mixtures of non Gaussian multivariate distributions. Such estimators will be applied in Face recognition.
Origin : Files produced by the author(s)
Loading...