Adversarial Debiasing with multiple sensitive attributes

[NiloyPurkait]

This request proposes changes to the original Adversarial debiasing algorithm by allowing the architecture to debias multiple sensitive attributes simultaneously. Code has been rewritten in tf 2.x

Note : Algorithm script included in algorithms.inprocessing directory. Jupyter notebook included under examples directory demonstrating use on COMPAS dataset.

Signed-off-by: Niloy Purkait [email protected]

[mnagired]

requesting a review from @hoffmansc, please take a look whenever you get a chance. thank you!