Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

tensorlow를 이용해 모델을 학습시키기 #1136

Open
ghost opened this issue Jul 27, 2021 · 4 comments
Open

tensorlow를 이용해 모델을 학습시키기 #1136

ghost opened this issue Jul 27, 2021 · 4 comments

Comments

@ghost
Copy link

ghost commented Jul 27, 2021

데이터 과학의 Tensorflow (python) 중에서 레몬에이드 판매예측 실습을 할때
모델을 많이 학습 시키면 어디까지 줄일 수 있을까에 대해 궁금증이 생겨 4만번 이상 학습을 시켜봤습니다
그런데 오히려 lose가 늘어나게 되었는데
왜 이런 결과가 나오게 되었을 까요?

@visiodeibc
Copy link
Collaborator

안녕하세요 hati00님!
테스트 데이터 세트에 너무 많이 트래이닝을 시킬 경우에는 해당 테스트 데이터 세트에 오버피팅이 일어날 수 있습니다.
학습된 모델이 너무 테스트 데이터 세트에 맞추어져서 실제 데이터로 트레이닝을 할 경우에
오히려 큰 loss를 가지게 되는 것이죠.

링크 여기서 좀 더 자세히 확인해 보실 수 있습니다!

@paullabkorea
Copy link

paullabkorea commented Jul 28, 2021

학습중에 일어나는 loss값 증가는 과소적합이나 과대적합 이슈는 아닙니다. 이 질문이 자주 나오는 질문이라 Tensorflow kr 커뮤니티의 글을 여기 첨부해드립니다.

포스팅 안에 있는 글도 천천히 읽어보시길 권해드립니다.

https://www.facebook.com/groups/TensorFlowKR/permalink/551967325144330/

@paullabkorea
Copy link

paullabkorea commented Jul 28, 2021

아, 제가 뎃글을 잘못 달았을 수도 있겠네요. 학습 중이 아니라 학습이 끝난 상태에서 테스트 데이터의 loss값을 얘기하셨던 것이었으면 위에 과대적합 이슈가 맞습니다.

@RayleighKim
Copy link
Collaborator

음, 역으로 질문을 드려야할 것 같습니다.
어떤 loss가 늘어난 걸까요? 트레이닝 데이터의 loss? 아니면 validation 데이터의? 혹은, 테스트데이터 일까요?
사용하신 코드나, 내용을 캡쳐해주시면 조금 더 상황파악에 도움이 될 것 같아요.

@progh2 progh2 added this to the 머신러닝 야학 3기 milestone Jul 30, 2021
@progh2 progh2 added ML / DL machine learning, deep learning 질문의 보완을 부탁드립니다 labels Jul 30, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

4 participants