Loss Function of Linear Regression, Logistic Regression, SVM

by allenlu2007

本文參考 Wiki SVM

Linear regression 的 loss function:

NewImage always >= 0

Error normal distribution 的 MLE estimation.  

Regularization linear regression 就是 MAP estimation.

 

Logistic regression 的 log-loss function:

 NewImage always >= 0

Error logistic distribution 的 MLE estimation.  

Regularization logistic regression 就是 MAP estimation.

 

SVM 的 loss function (hinge loss function)

NewImage always >= 0

Regularization SVM loss function 如下:

NewImage

Advertisements