Causal Inference in Machine Learning

← Back to document page

Causal models, revisited Instead of an exhaustive “table of interventional distributions”: G = (V, E), a causal graph with vertices V and edges E P( ), a probability over the “natural state” of V, parameterized by (G, ) is a causal model if pair (G, P) satisfies the Causal Markov condition

  Inference, Casual, Causal inference