Necessary and Sufficient Conditions for Causal Feature Selection in Time Series
             with Latent Common Causes

              Atalanti A. Mastakouri 1 Bernhard, Schlkopf 1 2 Dominik, Janzing 1

              Abstract               discovery of the causal graph from data is largely based on
    We study the identification of direct and indirect    the graphical criterion of d-separation formalizing the set
    causes on time series with latent variables, and     of conditional independen ...