navlie.batch.losses.L2Loss

class navlie.batch.losses.L2Loss

Bases: LossFunction

Standard L2 loss. Simply 0.5 * e * e , with the robust weight of ones.

loss(e: float)

The loss function defines the cost \(\rho(e)\), where \(e\) is an error term and is often a function of the design variable.

weight(x: float)

The weight that depends on the current value of the error. This is used to reweight our original nonlinear least squares problem.

previous

navlie.batch.losses.CauchyLoss

next

navlie.batch.losses.LossFunction