<content>

  • ML 전략
    • training loss
    • validation loss
    • test loss
    • real word loss
  • ML 모델의 목표를 설정하기 위한 방법

hyoeun-log.tistory.com/entry/WEEK5-ML-%EB%AA%A8%EB%8D%B8%EC%9D%98-%EB%AA%A9%ED%91%9C-%EC%84%A4%EC%A0%95%ED%95%98%EA%B3%A0-%EB%8B%AC%EC%84%B1%ED%95%98%EA%B8%B0

 

  • avoidable bias / variance의 비교를 통한 전략 세우기

hyoeun-log.tistory.com/entry/WEEK5-avoidable-bias-variance-%EB%B9%84%EA%B5%90%EB%A5%BC-%ED%86%B5%ED%95%9C-%EC%A0%84%EB%9E%B5-%EC%84%B8%EC%9A%B0%EA%B8%B0


Introuduction to ML strategy

 

현재 상황을 파악하고, 그에 맞게 대처하는 것이 중요하다.

 

  • 1. fit training set well on cost function
    • bigger network
    • Adam optimizer 사용
    • more epoch
  • 2. fit dev set well on cost function
    • regularization
    • more training data
  • 3. fit test set well on cost function
    • bigger dev set
  • 4. perform well in real word
    • change dev set
    • change cost function

 

 

 

+ Recent posts