Densely Connected Convolutional Networks: CVPR 2017 Best Paper Award Winner

Posted by ChaseT on Tuesday July 25, 2017 @10:16PM
from the arxiv.org dept.

( 3 comments | 29 points | on HN )

Re:Densely Connected Convolutional Networks: CVPR 2017 Best ... by GChevalier on Wednesday August 02, 2017 @12:56AM (#14908030)

Seems promising and easy to use in practice.

[ Parent | on HN ]

Re:Densely Connected Convolutional Networks: CVPR 2017 Best ... by Govindae on Wednesday July 26, 2017 @09:17AM (#14855941)

Am I understanding this correctly? The layers do not combine their inputs. Equation (2)

x_l = H_l([x_0, x_1, . . . , x_l−1])

Could be re-written: x_l = [H_l(x_0), H_l(x_1), . . . , H_l(x_l−1)]

[ Parent | on HN ]

Re:Densely Connected Convolutional Networks: CVPR 2017 Best ... by jordancampbell on Thursday July 27, 2017 @03:48PM (#14868765)

No, inputs to each layer are concatenated before being passed through H.

[ Parent | on HN ]