ML Paper Challenge Day 11— Deep Residual Learning for Image Recognition

Papers with Code - Deep Residual Learning for Image Recognition

Day 11: 2020.04.22
Paper: Deep Residual Learning for Image Recognition
Category: Model/CNN/Deep Learning/Image Recognition

Purpose: Tackle the problem of vanishing gradient in training

“When deeper networks are able to start converging, a degradation problem has been exposed: with the network depth increasing, accuracy gets saturated and then degrades rapidly. Unexpectedly, such degradation is not caused by overfitting, and adding more layers to a suitably deep model leads to higher training error.”

Solution: Residual Learning

Add a shortcut connection