全局样本损失函数 公式修正

全局样本损失函数 公式修正
This commit is contained in:
CoderOverflow 2019-04-03 16:58:14 +08:00 committed by GitHub
parent 0c4e637734
commit b886c60d29
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 1 additions and 1 deletions

View File

@ -488,7 +488,7 @@ cost(y,p(y|x))=-y\ln{p(y|x)-(1-y)\ln(1-p(y|x))}
$$
则全局样本损失函数为:
$$
cost(y,p(y|x)) = -\sum_{i=1}^m[y_i\ln p(y_i|x_i)-(1-y_i)\ln(1-p(y_i|x_i))]
cost(y,p(y|x)) = -\sum_{i=1}^m[y_i\ln p(y_i|x_i)+(1-y_i)\ln(1-p(y_i|x_i))]
$$
由此可看出,对数损失函数与极大似然估计的对数似然函数本质上是相同的。所以逻辑回归直接采用对数损失函数。