Convolutional case study

ke, 27 June 2018

[ deep_learning  ]
  • Classic networks:

LeNet-5

paper: [LeCun et al., 1998. Gradient-based learning applied to document recognition]

AlexNet

[Krizhevsky et al., 2012. ImageNet classification with deep convolutional neural networks]

VGG-16

[Simonyan & Zisserman 2015. Very deep convolutional networks for large-scale image recognigiton]

ResNet(layers 152)
[He et al., Deep residual networks for image recognition]

Residual block

Z[l+1] = W[l+1] * a[l] + b[l+1], a[l+1] = g(Z[l+1]), Z[l+2] = W[l+2] * a[l+1] + b[l+2], a[l+2]=g(Z[l+2])
Residual block: ……………………………………………………………., a[l+2]=g(Z[l+2]+a[l])

ResNet

1 * 1 convolutions

use 1 * 1 convolutions to shrink or expand the n_C channels

Inception network motivation
[Szegedy et al. 2014. Going deeper with convolutions]

Inception module

Inception network

  • Classic networks