Sorted L-One Penalized Estimation - theory and applications

Jeudi 20 avril 2017 15:30-16:30 - Malgorzata Bogdan - Uniwersytet Wroclawski - Instytut Matematyki

Résumé : Sorted L-One Penalized Estimation (SLOPE) is a relatively new method for selection of predictors and building predictive models based on the data from large data bases. I will present the general idea, current theoretical results and several new extensions aimed at different practical goals (including the analysis of graphical models and application for portfolio optimization).

Lieu : Salle 117-119

Sorted L-One Penalized Estimation - theory and applications  Version PDF