티스토리 뷰
Hypothesis Function의 정확도를 cost function으로 구할 수 있다. 대개, cost function 값이 낮을 수 록 정확도가 높다. Cost function의 형태에 따라 달라질 수 있다.
Mean Squared Error는 대표적인 cost function 중의 하나이다.
*이해되지 않는 설명
The mean is halved (1/2) as a convenience for the computation of the gradient descent, as the derivative term of the square function will cancel out the (1/2) term.
Gradient descent 계산할 때 도움이 되기 위해서 mean을 절반으로 나눈다. 제곱 함수를 미분하면 1/2는 의미가 없어진다.
'머신러닝' 카테고리의 다른 글
Random Forest (0) | 2021.12.29 |
---|---|
Skewed Class (0) | 2021.02.15 |
Batch gradient descent and Stochastic gradient descent (0) | 2021.01.10 |
Global minima and Local minima (0) | 2021.01.10 |
Categorical Variables (0) | 2021.01.08 |