An approach to non-adversarially minimize the symmetric Jeffreys divergence.
Can people collaborate to make ML fair?
