Add a causal representation learning example

[eb8680]

See"Causal Representation Learning from Multiple Distributions: A General Setting" for one description of the problem setup in the case of purely observational data. Implementing the approach in Section 4 of the paper and reproducing the toy experiments in Section 4.4 might be a good target for an example.

An alternative source for an example in the case of interventional data is "Nonparametric Identifiability of Causal Representations from Unknown Interventions".

[rpd29]

Would this still be useful to do? I read the second paper and am happy to take a crack at replicating it/creating an example using ChiRho

[eb8680]

Would this still be useful to do? I read the second paper and am happy to take a crack at replicating it/creating an example using ChiRho

Sure, though if you're specifically interested in this class of problems then I would suggest starting with the first paper because the setting is simpler and there's a very clear experimental description and set of results (specifically, figures 3&4) to reproduce.

[rpd29]

Gotcha, thank you!