Learning Causal Effects via Weighted Empirical Risk Minimization

Some slides inspired by those of Yonghan Jung at NeurIPS-20 Slides also inspired by Causality Shakil Jiwa Contents -Motivation -Identification and Estimation of Causal Effects -Empirical Risk Minimization -Paper Contribution -Algorithm Description -Experimental Results Motivation -Can a causal distribution be uniquely computed from a combination of the observational distribution P(V) and the causal graph G. -In general what do we do? Parametric model of conditional probabilities, suffers on computationally high dimensional data. -Effective estimators developed for when back door holds (known as ignorability in statistics) and further. Ie g-formula ?? P(y|do(x)) *in a way that we can estimate