ResNet----Identity Mappings

First, the structure

Second, the role

As the network continues to deepen, the effect on the training set decreases, and this is not caused by overfitting, because overfitting results in a good effect on the training set. By introducing identity mapping, if the network does not need deeper depth, it can walk the identity map of the road, and the residual mapping push to 0, so that the data is not forced through a deep network.

ResNet----Identity Mappings