smooth − labels smooth-labels smooth−L A B E L S ( flat slider superscript tab )
There are errors, omissions, and errors
in data labeling . You can use smooth labeling. If you think your data labeling is very good, you don’t need to
def smooth_labels(y_true, label_smoothing,num_classes):
return y_true * (1.0 - label_smoothing) + label_smoothing / num_classes