Seminar in applied mathematics and statistics

SPEAKER: Niklas Pfister (ETH Zurich, Department of Mathematics)

TITLE: Estimating the causal structure using different environmental settings without knowing them.

ABSTRACT: A recent promising method for causal structure learning is invariant causal prediction, which makes use of invariances inherent in causal models across different environmental settings. Unfortunately, this method requires that the environments are known, which is often not the case in practical data. In this talk, we will introduce an extension of invariant causal prediction, which instead of known environments uses a sequential ordering of the data. This occurs naturally in many data sets due to for example temporal collection methods.

-----

Tea and chocolate will be served on the 4th floor (04-4-19) after the seminar.

-----

UPCOMING SEMINARS:

February 3, 13.15: Stefan Bauer (ETH Zurich, Department of Computer Science)

 February 8, 15.15:  Boualem Djehiche (KTH Stockholm)