Speaker: Dongdong Ge

Title: Complexity and algorithms of some regression models

Abstract: In this talk we briefly investigate some sparse regression models
for large problems arising from signal processing, image reconstruction and
statistical data Processing. We proved that minimizing least square sum
problems with concave penalty functions, inlcuding bridge regression, is
strongly NP-Hard. We also provide efficient polynomial time algorithms to
find its appxomated KKT points.