Loading timeline…
20152010s
Kaiming He
Lead author of ResNet, whose residual connections made it possible to train networks hundreds of layers deep.
Organizations
Microsoft Research AsiaFacebook AI ResearchMassachusetts Institute of Technology
Major Achievements
- •Led the residual network (ResNet) work in 2015, winning ImageNet and becoming one of the most-cited papers in modern science.
- •Showed that skip connections solve the degradation problem that had capped useful network depth.
- •Co-developed Mask R-CNN for instance segmentation and Momentum Contrast for self-supervised vision.
- •Residual connections are now a standard component of the Transformer block.
Key Papers
- Deep Residual Learning for Image Recognition
Kaiming He et al. · CVPR 2016 · 2015
- Mask R-CNN
Kaiming He et al. · ICCV 2017