Introduction
Loss function (or cost function) is widely used for measuring the differences between predicted goal and real value. In different scenario, user should choose different loss function. This blog will introduce some markable loss function and their usage conditions and disadvantages.
Log Loss
Focal Loss
KL Divergence / Relative Entropy
Exponential Loss
Hinge Loss
Mean Square Error
Quadratic Loss
\lambda(x) = C (t-x)^2
Quadratic Loss Function – Wikipedia
Huber Loss / Smooth Mean Absolute Error
Log Cosh Loss
Quantile Loss
MSE
Mean Squared Error (MSE), also known as mean squared deviation (MSD) in statics.
MAE
Mean Absolute Error (MAE)
Mean Absolute Error – Wikipedia
Cross Entropy
Focal Loss
Introduction
References
Total References
Huber Loss和Focal Loss的原理与实现
Mean Square Error – Wikipedia
Quadratic Loss Function – Wikipedia
Mean Absolute Error – Wikipedia