Residual Network

A Convolutional Neural Network architecture.

Residual Blocks

Residual Blocks allows you to train much deeper networks.

Plain Network vs Residual Network

Helps with Vanishing and Exploding gradient problem. Helps in training much deeper networks.

Why doesn’t it hurt performance? It maintains identify function, which is essential just a single layer. While in plain networks, it’s hard to learn the identity function, which may hurt the performance.