BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Parameter identification in linear non-Gaussian causal models unde
 r general confounding - Mathias Drton (Technical University of Munich)
DTSTART:20260305T091500Z
DTEND:20260305T100000Z
UID:TALK244408@talks.cam.ac.uk
DESCRIPTION:Linear non-Gaussian causal models postulate that each random v
 ariable is a linear function of parent variables and non-Gaussian exogenou
 s error terms. We study identification of the linear coefficients when suc
 h models contain latent variables. Our focus is on the commonly studied ac
 yclic setting\, where each model corresponds to a directed acyclic graph (
 DAG). For this case\, prior literature has demonstrated that connections t
 o overcomplete independent component analysis yield effective criteria to 
 decide parameter identifiability in latent variable models. However\, this
  connection is based on the assumption that the observed variables linearl
 y depend on the latent variables. Departing from this assumption\, we trea
 t models that allow for arbitrary non-linear latent confounding. Our main 
 result is a graphical criterion that is necessary and sufficient for decid
 ing the generic identifiability of direct causal effects. Moreover\, we pr
 ovide an algorithmic implementation of the criterion with a run time that 
 is polynomial in the number of observed variables. Finally\, we report on 
 estimation heuristics based on the identification result and explore a gen
 eralization to models with feedback loops.\nJoint work with Daniele Tramon
 tano and Jalal Etesami.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR