For the same training procedure, a larger hidden size leads to a more compact decision boundary with large curva- ture. In the case of ERM training, the test ...
確定! 回上一頁