BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Constraint-based causal Discovery from NOnstationary/heterogeneous
  Data (CD-NOD) - Biwei Huang\, CMU
DTSTART:20180824T100000Z
DTEND:20180824T110000Z
UID:TALK109036@talks.cam.ac.uk
CONTACT:39846
DESCRIPTION:It is commonplace to encounter nonstationary or heterogeneous 
 data\, of which the under- lying generating process changes over time or a
 cross data sets. Such a distribution shift feature presents both challenge
 s and opportunities for causal discovery. In this talk\, I will present a 
 principled framework for causal discovery from such data\, called Constrai
 nt-based causal Discovery from NOnstationary/heterogeneous Data (CD-NOD)\,
  which addresses two important questions. First\, I will introduce an enha
 nced constraint-based procedure to detect variables whose local mechanisms
  change and recover the skeleton of the causal structure over observed var
 iables. Second\, I will present a way to determine causal orientations by 
 making use of independence changes in the data distribution implied by the
  underlying causal model\, benefiting from information carried by changing
  distributions. After learning the causal structure\, next\, I will discus
 s how to efficiently estimate the “driving force” of the nonstationari
 ty of a causal mechanism. That is\, we aim to extract from data a low-dime
 nsional and interpretable representation of the main components of the cha
 nges. Finally\, I will show that nonstationarity also benefits causal stru
 cture identification with stationary confounders.
LOCATION:dyson teaching room (TBC)
END:VEVENT
END:VCALENDAR