参照：
[stanford CS231n]Convolutional Neural Networks (CNNs / ConvNets)
[karpathy]ConvNetJS CIFAR-10 demo
【知乎译文】没有博士学位，照样玩转TensorFlow深度学习
【腾讯视频】TensorFlow and deep learning, without a PhD
Architecture Overview
- why ConvNet ?
Regular Neural Nets don’t scale well to full images. In CIFAR-10, images are only of size 32x32x3 (32 wide, 32 high, 3 color channels), so a single fully-connected neuron in a first hidden
layer of a regular Neural Network would have 32*32*3 = 3072 weights.
- 卷积层的作用？
local connectivity
spatial arrangement (Three hyperparameters control the size of the output volume: the depth, stride and zero-padding .)
parameter sharing
- 池化层的作用？
Its function is to progressively reduce the spatial size of the representation to reduce the amount of parameters and computation in the network, and hence to also control overfitting.