[D] Using CNN in Causal Inference

Hi,

I am new to Causal Inference, and recently I am trying to understand how machine learning could be used in the field.

I have been looking around and I don't seem to see works using CNN as the function approximator in the SCM formulation of causal model i.e. the f(X, noise) part, wouldn't that be useful for image data?

Please point out if my logic is flawed or this question doesn't make sense at all, thanks!

causal inference cnn function inference machine machine learning machinelearning noise part scm