AI学习笔记-CNN

Posted 2022-12-04 Dufre.WC

A toy ConvNet: X’s and O’s

Says whether a picture is of an X or an O

For example

Trickier Cases

• translation
• scaling
• rotation
• weight

Deciding is hard

What computers see?

The red area is incorrect

ConvNet match pieces of the image

Features match pieces of the image

Filtering(The math behind the match)

1. Line up the feature and the image patch
2. Multiply each image pixel by the corresponding feature pixel
3. Add them up
4. Divide by the total number of pixels in the feature
   

(output is average value)

Pooling(Shrinking the image stack)

1. Pick a window size (usually 2 or 3)
2. Pick a stride (usually 2)
3. Walk your window across your filtered images
4. From each window, take the maximum value
   Filter过后呢，数据量非常大，如果直接计算，计算量太大，非常耗时。比如对自动驾驶场景，人脸识别场景来说，必须在短时间内就做出反应。所以这里Pooling的作用就非常重要了。
   
   

Pooling layer
A stack of images becomes a stack of smaller images.

Normalization
Keep the math from breaking by tweaking each of the values just a bit.
Change everything negative to zero
ReLu layer
A stack of images becomes a stack of images with no negative values.

Layers get stacked
The output of one becomes the input of the next.

Deep stacking

Fully connected layer

Every value gets a vote

Vote depends on how strongly a value predicts X or O.

Backpropagation

Gradient descent

For each feature pixel and voting weight, adjust it up and down a bit and see how the error changes.
[

Putting it all together

A set of pixels becomes a set of votes

Hyperparameters(knobs)

(human set parameters)

• Convolution
  • Number of features
  • Size of features
• Pooling
  • Window size
  • Window stride
• Fully Connected
  • Number of neurons

Application

Image

Any 2D(or 3D) data
Things closer together are more closely related than things far away.

Sound

Text

Limitations

CNN only capture local “spatial” patterns in data.
If the data can’t be made to look like an image, CNN are less useful.
If your data is just as useful after swapping any of your columns with each other, then you can’t use Convolutional Neural Network.

Reference

• How Convolutional Neural Networks work？
• 大白话讲解卷积神经网络工作原理