BEGIN:VCALENDAR
VERSION:2.0
X-WR-CALNAME;VALUE=TEXT:Presenter- Victor Chernozhukov- Gaussian Approximations, Bootstrap, and Z-estimators when p >> n
PRODID:-//Harvard events data//EN
BEGIN:VEVENT
UID:event_297301_0
SUMMARY:Presenter- Victor Chernozhukov- Gaussian Approximations, Bootstrap, and Z-estimators when p >> n
DESCRIPTION:<p><span><strong><span>Author: </span></strong><span>Victor Chernozhukov</span></span></p><p><span><strong>Abstract:</strong> We show that central limit theorems hold for high-dimensional normalized means hitting high-dimensional rectangles. These results apply even when p&gt;&gt; n. These theorems provide Gaussian distributional approximations that are not pivotal, but they can be consistently estimated via Gaussian multiplier methods and the empirical bootstrap. These results are useful for building confidence bands and for multiple testing via the step-down methods. Moreover, these results hold for approximately linear estimators. As an application we show that these central limit theorems apply to normalized Z-estimators of p&gt; n target parameter in a class of problems, with estimating equations for each target parameter orthogonalized with respect to the nuisance functions being estimated via sparse methods. (This talk is based primarily on the joint work with Denis Chetverikov and Kengo Kato.)</span></p>
LOCATION:
STATUS:CONFIRMED
DTSTART:20141022T160000Z
DTEND:20141022T173000Z
END:VEVENT
END:VCALENDAR