Dat Phan

A Non-vacuous Test Error Guarantee for Deep Learning without Altering the Model

Non-vacuous Bounds for the test error of Deep Learning without any change to the trained models