Skip Connection Blocks

Skip Connection Blocks are building blocks for neural networks that feature skip connections. These skip connections 'skip' some layers allowing gradients to better flow through the network. Below you will find a continuously updating list of skip connection blocks:

METHOD YEAR PAPERS
Residual Block
2015 1082
Bottleneck Residual Block
2015 862
Dense Block
2016 201
Inverted Residual Block
2018 149
ResNeXt Block
2016 92
Non-Local Block
2017 42
ShuffleNet Block
2017 22
Wide Residual Block
2016 22
CBHG
2017 21
SRGAN Residual Block
2016 16
Inception-ResNet-v2-C
2016 10
Inception-ResNet-v2-B
2016 10
DV3 Convolution Block
2017 10
Pyramidal Residual Unit
2016 7
Reversible Residual Block
2017 7
Pyramidal Bottleneck Residual Unit
2017 6
DPN Block
2017 6
MelGAN Residual Block
2019 5
Dilated Bottleneck with Projection Block
2018 4
Dilated Bottleneck Block
2018 4
One-Shot Aggregation
2019 3
Two-Way Dense Layer
2018 3
SqueezeNeXt Block
2018 3
FBNet Block
2018 3
Selective Kernel
2019 3
Elastic Dense Block
2018 2
ParaNet Convolution Block
2019 2
OSA (identity mapping + eSE)
2019 2
Conditional DBlock
2019 2
GBlock
2019 2
CSPResNeXt Block
2019 2
Efficient Channel Attention
2019 2
DBlock
2019 2
DVD-GAN GBlock
2019 2
DVD-GAN DBlock
2019 2
Big-Little Module
2018 2
Res2Net Block
2019 2
Residual SRM
2019 1
NVAE Generative Residual Cell
2020 1
Global Context Block
2019 1
EESP
2018 1
Elastic ResNeXt Block
2018 1
NVAE Encoder Residual Cell
2020 1
Strided EESP
2018 1
Ghost Bottleneck
2019 1