2 citations · 2 across the 1 of their papers we have counts for
1 paper
Jeff Mitchell, Jeffrey S. Bowers
That shared features between train and test data are required for generalisation in artificial neural networks has been a common assumption of both proponents and critics of these…