ResNet (Residual Network)

Kaiming He, Xiangyu Zhang, Shaoqing Ren & Jian Sun, 2015

Introduced by He et al. in 2015, Residual Networks revolutionized deep learning by adding identity shortcut connections that skip one or more layers. Each residual block computes F(x) and adds the original input x, producing F(x)+x. This allows gradients to flow through the identity path during backpropagation, making it feasible to train networks hundreds of layers deep. The visualization shows data flowing through stacked residual blocks with skip connections arcing around each block.