文档介绍:Introduction Ns and
Sungjoon Choi
(sungjoon.******@)
CNN
2
Convolutional work
This is pretty much everything about the convolutional work.
Convolution + Subsampling + Full Connection
CNN
3
CNNs are basically layers of convolutions followed by subsampling and fully connected layers.
Intuitively speaking, convolutions and subsampling layers works as feature extraction layers while a fully connected layer classifies which category current input belongs to using extracted features.
Why so powerful?
4
https://starwarsanon./tag/darth-sidious-vs-yoda/
Local Invariance
Loosely speaking, as the convolution filters are ‘sliding’ over the input image, the exact location of the object we want to find does not matter much.
Compositionality
There is a hierarchy Ns. It is GOOD!
Convolution
5
/Feature_extraction_using_convolution
Get familiar with this
6
Zero-padding
Stride
Channel
Zero-padding
7
What is the size of the input?
What is the size of the output?
What is the size of the filter?
What is the size of the zero-padding?
Stride
8
Stride
9
(Left) Stride size: 1
(Right) Stride size: 2
If stride size equals the filter size, there will be no overlapping.
Conv2D
10
[filter_height, filter_width, in_channels, out_channels]
[batch, in_height=4, in_width=4, in_chnnel=3]
[filter_height=3, filter_width=3, in_channels=3, out_channels=7]
[batch, in_height, in_width, in_chnnel]