Seminario "On the Non-Asymptotic Properties of Regularized M-estimators"
Miércoles 29 de junio, 17h
Seminario de Economía
Paper Abstract
We propose a general framework for regularization in M-estimation problems under time dependent (absolutely regular-mixing) data which encompasses many of the existing estimators. We derive non-asymptotic concentration bounds for the regularized M-estimator. Our results exhibit a "variance-bias" trade-off, with the "variance" term being governed by a novel measure of the "size" of the parameter set. We also show that the mixing structure affect the variance term by scaling the number of observations; depending on the decay rate of the mixing coefficients, this scaling can even affect the asymptotic behavior. Finally, we propose a data-driven method for choosing the tuning parameters of the regularized estimator which yield the same (up to constants) concentration bound as one that optimally balances the "(squared) bias" and "variance" terms. We illustrate the results with several canonical examples of, both, non-parametric and high-dimensional models.
Demian Pouzo joined the Berkeley faculty in 2009 as an assistant professor after receiving his PhD in Economics from NYU. He also holds an MA and BA in Economics from Universidad Torcuato Di Tella (Argentina). Pouzo's current research interests include theoretical econometrics and macroeconomics.
Lugar: Campus Alcorta: Av. Figueroa Alcorta 7350, Ciudad de Buenos Aires.
Contacto: Cecilia Pilar Lafuente
Contacto: Cecilia Pilar Lafuente