Learning by Doing

Cross Entropy Loss with Softmax 미분에 대한 이해 본문

AI/cs231n

Cross Entropy Loss with Softmax 미분에 대한 이해

jh-rrr 2023. 7. 10. 20:09

https://madalinabuzau.github.io/2016/11/29/gradient-descent-on-a-softmax-cross-entropy-cost-function.html

 

Gradient descent on a Softmax cross-entropy cost function

In this blog post, you will learn how to implement gradient descent on a linear classifier with a Softmax cross-entropy loss function. I recently had to implement this from scratch, during the CS231 course offered by Stanford on visual recognition. Andrej

madalinabuzau.github.io

 

위 블로그에 정리된 증명을 통해 과정을 납득할 수 있었다.

아직 벡터(텐서)나 매트릭스의 미분에 익숙하지 않은 것 같아 관련 예제들을 집중하며 봐야겠다.

'AI > cs231n' 카테고리의 다른 글

Stanford CS224N - NLP w/ DL | Winter 2021 | Lecture 4 - Syntactic Structure and Dependency Parsing  (0) 2022.09.16
'AI/cs231n' Related Articles more