Search⌘ K

ResNet Block

Learn how ResNet blocks use residual learning to ease training of deep networks by enabling identity mappings. Understand the structure of these blocks with convolution layers and practice coding a ResNet block function to improve image classification models.

Chapter Goals:

  • Understand the purpose of residual learning
  • Learn the structure of a ResNet building block
  • Create a function for a ResNet building block

A. Learning the identity

In the previous chapter, we touched upon residual learning, i.e. the process of a block learning the residual function:

FB(x)=HB(x)x\small \mathcal{F}_B(\mathbf{x}) = \mathcal{H}_B(\mathbf{x}) - \mathbf{x} ...