News for hackers. Stuff that matters.
Re:The many Shapley values for model explanation by rficcaglia on Saturday August 24, 2019 @10:42AM (#20787373)
> This is the crux of the matter: In taking the model seriously (even "literally"), BS evaluates it on points in feature space that may never occur in practice, and may not even be realizable.

Is it April 1?