← Voltar para Pesquisa
Deep Residual Learning for Image Recognition
He et al. • 2015
VisionDeep LearningCNN
Resumo
ResNet introduced 'skip connections' (residual learning), allowing neural networks to be trained at unprecedented depths (hundreds of layers) without vanishing gradients. This architecture became the default for computer vision for years.
Por Que Importa
- Enabled training of very deep networks
- Winner of ILSVRC 2015
- Skip connections are now standard across architectures
Perguntar sobre este artigo
Loading chat...
Deep Residual Learning for Image Recognition
He et al. • 2015
VisionDeep LearningCNN
Resumo
ResNet introduced 'skip connections' (residual learning), allowing neural networks to be trained at unprecedented depths (hundreds of layers) without vanishing gradients. This architecture became the default for computer vision for years.
Por Que Importa
- Enabled training of very deep networks
- Winner of ILSVRC 2015
- Skip connections are now standard across architectures
Perguntar sobre este artigo
Loading chat...