BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:"Statistical Estimation with Random Forests"
DTSTART:20160129T140000
DTEND:20160129T150000
DTSTAMP:20260406T224459Z
UID:12566617de2931dced2c4042a51811a68a190b044a7349eab523e05f
CATEGORIES:Conferences - Seminars
DESCRIPTION:Dr. Stefan WAGER (Stanford University)\nRandom forests\, intro
 duced by Breiman (2001)\, are among the most widely used machine learning 
 algorithms today\, with applications in fields as varied as ecology\, gene
 tics\, and remote sensing. Random forests have been found empirically to f
 it complex interactions in high dimensions\, all while remaining strikingl
 y resilient to overfitting. In principle\, these qualities also ought to m
 ake random forests good statistical estimators. However\, our current unde
 rstanding of the statistics of random forest predictions is not good enoug
 h to make random forests usable as a part of a standard applied statistics
  pipeline: in particular\, we lack robust consistency guarantees and asymp
 totic inferential tools. In this talk\, I will present some recent results
  that seek to overcome these limitations. The first half of the talk devel
 ops a Gaussian theory for random forests in low dimensions that allows for
  valid asymptotic inference\, and applies the resulting methodology to the
  problem of heterogeneous treatment effect estimation. The second half of 
 the talk then considers high-dimensional properties of regression trees an
 d forests in a setting motivated by the work of Berk et al. (2013) on vali
 d post-selection inference: at a high level\, we find that the amount by w
 hich a random forest can overfit to training data scales only logarithmica
 lly in the ambient dimension of the problem.\nThis talk is based on joint 
 work with Susan Athey\, Bradley Efron\, Trevor Hastie\, and Guenther Walth
 er.
LOCATION:CM 1 5 https://plan.epfl.ch/?room==CM%201%205
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR