CNN, Convolution, Kernel, Pooling 이미지들

이런 그림들 너무 좋다..🥰

 

Convoluting a 5x5x1 image with a 3x3x1 kernel to get a 3x3x1 convolved feature

 

Convolution Operation with Stride Length = 2

 

 

SAME padding:  5x5x1 image is padded with 0s to create a 6x6x1 image

 

 

 

 

 

 

합성곱(合成-), 또는 콘벌루션(convolution)은 

하나의 함수와 또 다른 함수를 반전 이동한 값을 곱한 다음, 

구간에 대해 적분하여 새로운 함수를 구하는 수학 연산자

 

 

3x3 pooling over 5x5 convolved feature

 

Type of Pooling

 

 

 

Pooling으로 Overfitting을 방지한다고 한다. 

 

 

 

 

 

 

A CNN sequence to classify handwritten digits

 

 

 

 

출처 : https://hobinjeong.medium.com/cnn%EC%97%90%EC%84%9C-pooling%EC%9D%B4%EB%9E%80-c4e01aa83c83

 

 

 

 

출처: https://brunch.co.kr/@gdhan/7

 

 

 

 

 

출처: https://www.analyticsvidhya.com/blog/2022/01/convolutional-neural-network-an-overview/

 

 

 

 

 

 

출처: https://learnopencv.com/understanding-convolutional-neural-networks-cnn/

 

 

 

 

 

출처: https://velog.io/@svenskpotatis/%EB%94%A5%EB%9F%AC%EB%8B%9D-CNN

 

 

 

 

 

 

 

 

 

 

 

이미지들 메인 출처! 출처 없는 이미지는 이곳이 출처:

https://towardsdatascience.com/a-comprehensive-guide-to-convolutional-neural-networks-the-eli5-way-3bd2b1164a53

A Comprehensive Guide to Convolutional Neural Networks — the ELI5 way

 

일단 보려고.. 킵

https://velog.io/@kim_haesol/CNN-%EA%B8%B0%EC%B4%88%EC%84%A4%EB%AA%85