Why each block in Deep residual network has two convolutional layers instead of one?

Question

This figure shows a basic block of a residual network. What it has two convolutional layers? What will happen when it has only one convolutional layer?

score 1 · Answer 1 · answered Mar 15 '18 at 07:53

1

It should still work with somewhat similar results. There are no problems with skipping only one layer.

See https://openreview.net/forum?id=HkwBEMWCZ and densenets https://arxiv.org/abs/1608.06993

answered Mar 15 '18 at 07:53

Umang Gupta

15,022
6
48
66

Why each block in Deep residual network has two convolutional layers instead of one?

1 Answers1