Logistic Regression
Logistic regression is a classifier despite its name. Binary Cross Entropy with logits is used because MSE and MAE are regression losses. Logistic regression applies a sigmoid activation to a linear model output to map values into probabil…
1 sources - 6 claims
Logistic regression is a classifier despite its name. Binary Cross Entropy with logits is used because MSE and MAE are regression losses. Logistic regression applies a sigmoid activation to a linear model output to map values into probability space. BCE with logits combines sigmoid activation and cross entropy in a numerically stable operation.