Open Problem: How fast can a multiclass test set be overfit?

[edit]

Vitaly Feldman, Roy Frostig, Moritz Hardt ;
Proceedings of the Thirty-Second Conference on Learning Theory, PMLR 99:3185-3189, 2019.

Abstract

We ask how many measurements of the accuracy on a multiclass benchmark are needed to achieve a given amount of overfitting.

Related Material