On the Variance of Neural Network Training with respect to Test Sets and Distributions.

Keller Jordan

Browse the full ICLR paper archive.