AI Hub
All papers
Architecture

Deep Residual Learning for Image Recognition

Microsoft·December 10, 2015

Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun

View on arXiv

TL;DR

Introduces residual (skip) connections, enabling networks hundreds of layers deep to train.

Why it matters

Residual connections are now everywhere, including inside every Transformer block.

Related terms