Package org.ddogleg.optimization.loss
package org.ddogleg.optimization.loss
-
ClassDescriptionLoss function inspired by the Cauchy distribution, a.k.a Lorentzian loss function.Implementation of the smooth Cauchy loss functionImplementation of the smooth Cauchy loss gradientResidual loss function for regression.Base class for loss functions.Analytical gradient for a function that implements
LossFunction.Huber Loss is a robust loss function that is less sensitive to outliers than the squared error loss.Implementation of the Huber loss functionImplementation of the Huber Loss gradientSmooth approximation to the huber loss [1].Implementation of the smooth Huber loss functionImplementation of the smooth Huber loss gradientThis loss function simply passes through the computed "residual".Iteratively Reweighted Least-Squares (IRLS) allows the weights to be recomputed every iteration.Squared error loss function.Tukey loss (Tukey's biweight function) has similar behavior toLossHuberbut is less sensitive to outliers because they contribute nothing to the loss.Implementation of the Tukey loss functionImplementation of the Tukey Loss gradientA weighted least squares cost function.