1 paper
Itamar Harel, William M. Hoza, Gal Vardi +3
We study the overfitting behavior of fully connected deep Neural Networks (NNs) with binary weights fitted to perfectly classify a noisy training set. We consider interpolation usi…