zoom link, opens an external URL in a new window

Meeting-ID: 280 519 2121
Passwort: 584190

Abstract:

The classical latent factor model for (generalised) ill-posed linear regression is extended by assuming that, up to an unknown orthogonal transformation, the features consist of subsets that are relevant and irrelevant to the response. Furthermore, a joint low-dimensionality is imposed only on the relevant features and the response variable. This framework not only allows for a comprehensive study of the partial-least-squares (PLS) algorithm under random design, but also sheds light on the performance of other regularisation methods that exploit sparsity or unsupervised projection. Moreover, we propose a novel iteratively-reweighted-partial-least-squares (IRPLS) algorithm for ill-posed generalised linear models and obtain its convergence rates working in the suggested framework. This is a joint work with Gianluca Finocchio.